Backpropagation applied to handwritten zip code recognition
Introduction to Artificial Intelli智慧树知到课后章节答案2023年
Introduction to Artificial Intelligence智慧树知到课后章节答案2023年下哈尔滨工程大学哈尔滨工程大学第一章测试1.All life has intelligence The following statements about intelligence arewrong（）A:All life has intelligence B:Bacteria do not have intelligence C:At present,human intelligence is the highest level of nature D:From the perspective of life, intelligence is the basic ability of life to adapt to the natural world答案:Bacteria do not have intelligence2.Which of the following techniques is unsupervised learning in artificialintelligence?（）A:Neural network B:Support vector machine C:Decision tree D:Clustering答案:Clustering3.To which period can the history of the development of artificial intelligencebe traced back?（）A:1970s B:Late 19th century C:Early 21st century D:1950s答案:Late 19th century4.Which of the following fields does not belong to the scope of artificialintelligence application?（）A:Aviation B:Medical C:Agriculture D:Finance答案:Aviation5.The first artificial neuron model in human history was the MP model,proposed by Hebb.（）A:对 B:错答案:错6.Big data will bring considerable value in government public services, medicalservices, retail, manufacturing, and personal location services. （）A:错 B:对答案:对第二章测试1.Which of the following options is not human reason:（）A:Value rationality B:Intellectual rationality C:Methodological rationalityD:Cognitive rationality答案:Intellectual rationality2.When did life begin? （）A:Between 10 billion and 4.5 billion years B:Between 13.8 billion years and10 billion years C:Between 4.5 billion and 3.5 billion years D:Before 13.8billion years答案:Between 4.5 billion and 3.5 billion years3.Which of the following statements is true regarding the philosophicalthinking about artificial intelligence?（）A:Philosophical thinking has hindered the progress of artificial intelligence.B:Philosophical thinking has contributed to the development of artificialintelligence. C:Philosophical thinking is only concerned with the ethicalimplications of artificial intelligence. D:Philosophical thinking has no impact on the development of artificial intelligence.答案:Philosophical thinking has contributed to the development ofartificial intelligence.4.What is the rational nature of artificial intelligence?（）A:The ability to communicate effectively with humans. B:The ability to feel emotions and express creativity. C:The ability to reason and make logicaldeductions. D:The ability to learn from experience and adapt to newsituations.答案:The ability to reason and make logical deductions.5.Which of the following statements is true regarding the rational nature ofartificial intelligence?（）A:The rational nature of artificial intelligence includes emotional intelligence.B:The rational nature of artificial intelligence is limited to logical reasoning.C:The rational nature of artificial intelligence is not important for itsdevelopment. D:The rational nature of artificial intelligence is only concerned with mathematical calculations.答案:The rational nature of artificial intelligence is limited to logicalreasoning.6.Connectionism believes that the basic element of human thinking is symbol,not neuron; Human's cognitive process is a self-organization process ofsymbol operation rather than weight. （）A:对 B:错答案:错第三章测试1.The brain of all organisms can be divided into three primitive parts:forebrain, midbrain and hindbrain. Specifically, the human brain is composed of brainstem, cerebellum and brain (forebrain). （）A:错 B:对答案:对2.The neural connections in the brain are chaotic. （）A:对 B:错答案:错3.The following statement about the left and right half of the brain and itsfunction is wrong （）.A:When dictating questions, the left brain is responsible for logical thinking,and the right brain is responsible for language description. B:The left brain is like a scientist, good at abstract thinking and complex calculation, but lacking rich emotion. C:The right brain is like an artist, creative in music, art andother artistic activities, and rich in emotion D:The left and right hemispheres of the brain have the same shape, but their functions are quite different. They are generally called the left brain and the right brain respectively.答案:When dictating questions, the left brain is responsible for logicalthinking, and the right brain is responsible for language description.4.What is the basic unit of the nervous system?（）A:Neuron B:Gene C:Atom D:Molecule答案:Neuron5.What is the role of the prefrontal cortex in cognitive functions?（）A:It is responsible for sensory processing. B:It is involved in emotionalprocessing. C:It is responsible for higher-level cognitive functions. D:It isinvolved in motor control.答案:It is responsible for higher-level cognitive functions.6.What is the definition of intelligence?（）A:The ability to communicate effectively. B:The ability to perform physicaltasks. C:The ability to acquire and apply knowledge and skills. D:The abilityto regulate emotions.答案:The ability to acquire and apply knowledge and skills.第四章测试1.The forward propagation neural network is based on the mathematicalmodel of neurons and is composed of neurons connected together by specific connection methods. Different artificial neural networks generally havedifferent structures, but the basis is still the mathematical model of neurons.（）A:对 B:错答案:对2.In the perceptron, the weights are adjusted by learning so that the networkcan get the desired output for any input. （）A:对 B:错答案:对3.Convolution neural network is a feedforward neural network, which hasmany advantages and has excellent performance for large image processing.Among the following options, the advantage of convolution neural network is（）.A:Implicit learning avoids explicit feature extraction B:Weight sharingC:Translation invariance D:Strong robustness答案:Implicit learning avoids explicit feature extraction;Weightsharing;Strong robustness4.In a feedforward neural network, information travels in which direction?（）A:Forward B:Both A and B C:None of the above D:Backward答案:Forward5.What is the main feature of a convolutional neural network?（）A:They are used for speech recognition. B:They are used for natural languageprocessing. C:They are used for reinforcement learning. D:They are used forimage recognition.答案:They are used for image recognition.6.Which of the following is a characteristic of deep neural networks?（）A:They require less training data than shallow neural networks. B:They havefewer hidden layers than shallow neural networks. C:They have loweraccuracy than shallow neural networks. D:They are more computationallyexpensive than shallow neural networks.答案:They are more computationally expensive than shallow neuralnetworks.第五章测试1.Machine learning refers to how the computer simulates or realizes humanlearning behavior to obtain new knowledge or skills, and reorganizes the existing knowledge structure to continuously improve its own performance.（）A:对 B:错答案:对2.The best decision sequence of Markov decision process is solved by Bellmanequation, and the value of each state is determined not only by the current state but also by the later state.（）A:对 B:错答案:对3.Alex Net's contributions to this work include: （）.A:Use GPUNVIDIAGTX580 to reduce the training time B:Use the modified linear unit (Re LU) as the nonlinear activation function C:Cover the larger pool to avoid the average effect of average pool D:Use the Dropouttechnology to selectively ignore the single neuron during training to avoid over-fitting the model答案:Use GPUNVIDIAGTX580 to reduce the training time;Use themodified linear unit (Re LU) as the nonlinear activation function;Cover the larger pool to avoid the average effect of average pool;Use theDropout technology to selectively ignore the single neuron duringtraining to avoid over-fitting the model4.In supervised learning, what is the role of the labeled data?（）A:To evaluate the model B:To train the model C:None of the above D:To test the model答案:To train the model5.In reinforcement learning, what is the goal of the agent?（）A:To identify patterns in input data B:To minimize the error between thepredicted and actual output C:To maximize the reward obtained from theenvironment D:To classify input data into different categories答案:To maximize the reward obtained from the environment6.Which of the following is a characteristic of transfer learning?（）A:It can only be used for supervised learning tasks B:It requires a largeamount of labeled data C:It involves transferring knowledge from onedomain to another D:It is only applicable to small-scale problems答案:It involves transferring knowledge from one domain to another第六章测试1.Image segmentation is the technology and process of dividing an image intoseveral specific regions with unique properties and proposing objects ofinterest. In the following statement about image segmentation algorithm, the error is （）.A:Region growth method is to complete the segmentation by calculating the mean vector of the offset. B:Watershed algorithm, MeanShift segmentation,region growth and Ostu threshold segmentation can complete imagesegmentation. C:Watershed algorithm is often used to segment the objectsconnected in the image. D:Otsu threshold segmentation, also known as themaximum between-class difference method, realizes the automatic selection of global threshold T by counting the histogram characteristics of the entire image答案:Region growth method is to complete the segmentation bycalculating the mean vector of the offset.2.Camera calibration is a key step when using machine vision to measureobjects. Its calibration accuracy will directly affect the measurementaccuracy. Among them, camera calibration generally involves the mutualconversion of object point coordinates in several coordinate systems. So,what coordinate systems do you mean by "several coordinate systems" here?（）A:Image coordinate system B:Image plane coordinate system C:Cameracoordinate system D:World coordinate system答案:Image coordinate system;Image plane coordinate system;Camera coordinate system;World coordinate systemmonly used digital image filtering methods:（）.A:bilateral filtering B:median filter C:mean filtering D:Gaussian filter答案:bilateral filtering;median filter;mean filtering;Gaussian filter4.Application areas of digital image processing include:（）A:Industrial inspection B:Biomedical Science C:Scenario simulation D:remote sensing答案:Industrial inspection;Biomedical Science5.Image segmentation is the technology and process of dividing an image intoseveral specific regions with unique properties and proposing objects ofinterest. In the following statement about image segmentation algorithm, the error is ( ).A:Otsu threshold segmentation, also known as the maximum between-class difference method, realizes the automatic selection of global threshold T by counting the histogram characteristics of the entire imageB: Watershed algorithm is often used to segment the objects connected in the image. C:Region growth method is to complete the segmentation bycalculating the mean vector of the offset. D:Watershed algorithm, MeanShift segmentation, region growth and Ostu threshold segmentation can complete image segmentation.答案:Region growth method is to complete the segmentation bycalculating the mean vector of the offset.第七章测试1.Blind search can be applied to many different search problems, but it has notbeen widely used due to its low efficiency.（）A:错 B:对答案:对2.Which of the following search methods uses a FIFO queue （）.A:width-first search B:random search C:depth-first search D:generation-test method答案:width-first search3.What causes the complexity of the semantic network （）.A:There is no recognized formal representation system B:The quantifiernetwork is inadequate C:The means of knowledge representation are diverse D:The relationship between nodes can be linear, nonlinear, or even recursive 答案:The means of knowledge representation are diverse;Therelationship between nodes can be linear, nonlinear, or even recursive4.In the knowledge graph taking Leonardo da Vinci as an example, the entity ofthe character represents a node, and the relationship between the artist and the character represents an edge. Search is the process of finding the actionsequence of an intelligent system.（）A:对 B:错答案:对5.Which of the following statements about common methods of path search iswrong（）A:When using the artificial potential field method, when there are someobstacles in any distance around the target point, it is easy to cause the path to be unreachable B:The A* algorithm occupies too much memory during the search, the search efficiency is reduced, and the optimal result cannot beguaranteed C:The artificial potential field method can quickly search for acollision-free path with strong flexibility D:A* algorithm can solve theshortest path of state space search答案:When using the artificial potential field method, when there aresome obstacles in any distance around the target point, it is easy tocause the path to be unreachable第八章测试1.The language, spoken language, written language, sign language and Pythonlanguage of human communication are all natural languages.（）A:对 B:错答案:错2.The following statement about machine translation is wrong （）.A:The analysis stage of machine translation is mainly lexical analysis andpragmatic analysis B:The essence of machine translation is the discovery and application of bilingual translation laws. C:The four stages of machinetranslation are retrieval, analysis, conversion and generation. D:At present,natural language machine translation generally takes sentences as thetranslation unit.答案:The analysis stage of machine translation is mainly lexical analysis and pragmatic analysis3.Which of the following fields does machine translation belong to? （）A:Expert system B:Machine learning C:Human sensory simulation D:Natural language system答案:Natural language system4.The following statements about language are wrong: （）。
2024年牛津版初一上学期期中英语试卷与参考答案
2024年牛津版英语初一上学期期中模拟试卷与参考答案一、听力部分（本大题有20小题，每小题1分，共20分）1、听力原文：W: Good morning, Class 1. I hope you’re all ready for the mid-term exam. Let’s begin with the first question. Please listen carefully.Question: How many subjects are there in the school’s curriculum?A)3B)4C)5D)6Answer: C解析：在听力原文中，教师提到“Let’s begin with the first question”，随后会提到课程数量，答案为“There are 5 subjects in our curriculum.” 因此，正确答案是 C。
2、听力原文：M: Hey, Lily, how was your weekend?W: Well, I went to the library to study for the math exam. I was there for about 3 hours. How about you?Question: How long did Lily spend at the library?A) 1 hourB) 2 hoursC) 3 hoursD) 4 hoursAnswer: C解析：在听力对话中，Lily提到她在图书馆学习了大约 3 个小时。
因此，正确答案是 C。
3、Listen to the following conversation and choose the best answer to the question you hear.W: Hi, John. How was your weekend?M: It was pretty good, thanks. I went hiking with my friends.W: That sounds fun. Did you see any wildlife?M: Yeah, we saw a few deer and some birds. It was amazing.Q: What activity did John do over the weekend?A. He went hiking.B. He watched movies.C. He went shopping.D. He visited a museum.Answer: AExplanation: In the conversation, John mentions that he “went hiking with my friends,” which indicates that he went hiking over the weekend.4、Listen to the following dialogue and answer the question.M: Hi, Lisa. How was your math test yesterday?W: Oh, it was really tough. There were a lot of problems I didn’t understand.M: I know, right? I had trouble with the geometry questions too.W: Yeah, and the algebra section was confusing. Do you think we’ll get the answers back soon?M: I hope so. I want to know where I went wrong.Q: What subject did Lisa have trouble with in the test?A. EnglishB. ScienceC. MathD. HistoryAnswer: CExplanation: Lisa explicitly states that she had trouble with the “math test,” which is the subject she struggled with.5.You are listening to a conversation between a student and a teacher. The student is asking for advice on choosing a subject for his project. Listen carefully and answer the question below.Question: What subject does the student want to choose for his project?A. HistoryB. ScienceC. ArtD. MusicAnswer: CExplanation: The student mentions that he has a strong interest in art and wants to explore it further, so he is interested in choosing art as his project subject.6.You are listening to a short dialogue between two friends discussing their weekend plans. Listen carefully and answer the question below.Question: What is the friend’s plan for the weekend?A. Visiting a museumB. Going to the moviesC. Attending a sports eventD. Staying home and readingAnswer: AExplanation: The friend says, “I think I’ll go to the museum this weekend,” which indicates that visiting a museum is their plan.7、You hear a conversation between two students, Tom and Lisa, discussing their weekend plans. Listen to the conversation and answer the question.Question: What activity does Lisa plan to do on Saturday afternoon?A. Visit the libraryB. Go to the cinemaC. Practice pianoAnswer: CExplanation: In the conversation, Lisa mentions, “Well, I’m planning to practice the piano on Saturday afternoon.” Therefore, the correct answer is C. Practice piano.8、You hear a weather report for Oxford. Listen to the report and answer the question.Question: What is the temperature forecast for today in Oxford?A. 15°CB. 10°CC. 20°CAnswer: BExplanation: The weather report st ates, “Today’s forecast for Oxford is a mild10 degrees Celsius with a few scattered clouds.” Hence, the correct answer isB. 10°C.9.You hear a conversation between two students, Alice and Ben, discussing their weekend plans.Alice: “I’m thinking of going hiking this weekend. Do you want to come with me?”Ben: “That sounds fun! But I’m not sure if I’m up for the challenge.”Question: What does Ben mean when he says “I’m not sure if I’m up for the challenge”?A)He is confident he can hike.B)He is worried about his fitness.C)He doesn’t like hiking.D)He wants to go hiking but can’t.Answer: B) He is worried about his fitness.解析：Ben在回答中使用了”I’m not sure if I’m up for the challenge”这句话，表明他对自己的体能是否能够应对徒步挑战感到不确定，因此答案是B。
chatgpt 浓缩文献综述的指令
文献综述是科学研究中非常重要的一环，通过对已有学术文献的整理、归纳和总结，可以为后续研究提供重要的参考和指导。
在计算机科学领域，自然语言处理是一个备受关注的研究方向，而chatgpt作为一种经典的人工智能模型，其在自然语言处理领域的应用备受关注。
chatgpt是由Open本人开发的一种基于深度学习的对话生成模型，其可以生成接近人类水平的自然语言对话。
下面将通过浓缩文献综述的形式，对chatgpt在自然语言处理领域的相关研究进行梳理和总结。
1. chatgpt的基本原理chatgpt是基于Transformer模型的改进版本，其核心原理是通过对大规模文本语料进行预训练，学习文本中的语言模式和语义信息，从而达到生成流畅、连贯对话的目的。
模型的训练采用了自监督学习的方法，通过最大化文本序列的联合概率来优化模型参数，使得模型可以对输入的自然语言进行理解和生成。
在具体的应用中，chatgpt可以用于对话生成、文本摘要、问答系统等多种自然语言处理任务。
2. chatgpt的发展历程chatgpt的发展经历了多个版本的迭代，从最初的GPT-1到目前比较成熟的GPT-3，模型的规模和性能都得到了显著的提升。
随着模型规模的不断扩大和训练数据的不断增加，chatgpt在自然语言处理领域的表现也逐渐趋近甚至超越了人类水平，成为了当前最受关注的人工智能模型之一。
3. chatgpt在对话生成领域的应用chatgpt在对话生成方面具有非常广泛的应用，包括智能客服、聊天机器人、虚拟助手等。
通过与用户进行对话交互，chatgpt可以实现智能问答、情感分析、任务指导等多种功能，极大地丰富了人机交互的方式，改变了人们日常生活和工作中的沟通方式。
4. chatgpt在文本摘要领域的应用文本摘要是自然语言处理领域的一个重要任务，其旨在从文本中提取出最重要的信息，生成简洁、精炼的摘要内容。
chatgpt可以通过对输入文本进行理解和归纳，自动生成符合人类习惯的文本摘要，极大地提高了文本处理效率和用户体验。
Neural networks and deep learning 1
The human visual system is one of the wonders of the world.Consider the following sequence of handwritten digits:Most people effortlessly recognize those digits as 504192. That ease is deceptive. In each hemisphere of our brain, humans have aprimary visual cortex, also known as V1, containing 140 millionneurons, with tens of billions of connections between them. And yet human vision involves not just V1, but an entire series of visual cortices - V2, V3, V4, and V5 - doing progressively more compleximage processing. We carry in our heads a supercomputer, tunedby evolution over hundreds of millions of years, and superbly adapted to understand the visual world. Recognizing handwrittendigits isn't easy. Rather, we humans are stupendously,astoundingly good at making sense of what our eyes show us. Butnearly all that work is done unconsciously. And so we don't usually appreciate how tough a problem our visual systems solve.The difficulty of visual pattern recognition becomes apparent ifyou attempt to write a computer program to recognize digits likethose above. What seems easy when we do it ourselves suddenly becomes extremely difficult. Simple intuitions about how we recognize shapes - "a 9 has a loop at the top, and a vertical stroke inthe bottom right" - turn out to be not so simple to expressalgorithmically. When you try to make such rules precise, you quickly get lost in a morass of exceptions and caveats and specialcases. It seems hopeless.Neural networks approach the problem in a different way. The ideais to take a large number of handwritten digits, known as trainingexamples,CHAPTER 1Using neural nets to recognize handwritten digits Neural Networks and Deep Learning What this book is about On the exercises and problems Using neural nets to recognize handwritten digits How the backpropagation algorithm works Improving the way neural networks learn A visual proof that neural nets can compute any function Why are deep neural networks hard to train?Deep learning Acknowledgements Frequently Asked Questions Sponsors Thanks to all the supporters who made the book possible. Thanks also to all the contributors to the Bugfinder Hall of Fame .The book is currently a beta release,and is still under active development. Please send error reports to mn@.For other enquiries, please see the FAQ first.Resources Code repository Mailing list for book announcementsMichael Nielsen's projectannouncement mailing listand then develop a system which can learn from those trainingexamples. In other words, the neural network uses the examples to automatically infer rules for recognizing handwritten digits.Furthermore, by increasing the number of training examples, the network can learn more about handwriting, and so improve its accuracy. So while I've shown just 100 training digits above,perhaps we could build a better handwriting recognizer by using thousands or even millions or billions of training examples.In this chapter we'll write a computer program implementing a neural network that learns to recognize handwritten digits. The program is just 74 lines long, and uses no special neural network libraries. But this short program can recognize digits with an accuracy over 96 percent, without human intervention.Furthermore, in later chapters we'll develop ideas which canimprove accuracy to over 99 percent. In fact, the best commercial neural networks are now so good that they are used by banks to process cheques, and by post offices to recognize addresses.We're focusing on handwriting recognition because it's anexcellent prototype problem for learning about neural networks in general. As a prototype it hits a sweet spot: it's challenging - it's no small feat to recognize handwritten digits - but it's not so difficult as to require an extremely complicated solution, or tremendous computational power. Furthermore, it's a great way to develop more advanced techniques, such as deep learning. And sothroughout the book we'll return repeatedly to the problem ofhandwriting recognition. Later in the book, we'll discuss how these ideas may be applied to other problems in computer vision, and By Michael Nielsen / Dec 2014Typesetting math: 90%also in speech, natural language processing, and other domains.Of course, if the point of the chapter was only to write a computer program to recognize handwritten digits, then the chapter would be much shorter! But along the way we'll develop many key ideas about neural networks, including two important types of artificial neuron (the perceptron and the sigmoid neuron), and the standard learning algorithm for neural networks, known as stochasticgradient descent. Throughout, I focus on explaining why things are done the way they are, and on building your neural networks intuition. That requires a lengthier discussion than if I justpresented the basic mechanics of what's going on, but it's worth it for the deeper understanding you'll attain. Amongst the payoffs, by the end of the chapter we'll be in position to understand what deep learning is, and why it matters.PerceptronsWhat is a neural network? To get started, I'll explain a type of artificial neuron called a perceptron . Perceptrons were developed in the 1950s and 1960s by the scientist Frank Rosenblatt , inspired by earlier work by Warren McCulloch and Walter Pitts . Today, it's more common to use other models of artificial neurons - in this book, and in much modern work on neural networks, the main neuron model used is one called the sigmoid neuron . We'll get to sigmoid neurons shortly. But to understand why sigmoid neurons are defined the way they are, it's worth taking the time to first understand perceptrons.So how do perceptrons work? A perceptron takes several binary inputs,, and produces a single binary output:In the example shown the perceptron has three inputs, .In general it could have more or fewer inputs. Rosenblatt proposed a simple rule to compute the output. He introduced weights , , real numbers expressing the importance of the respective inputs to the output. The neuron's output, or , is,,…x 1x 2,,x 1x 2x 3,,…w 1w 201determined by whether the weighted sum is less than or greater than some threshold value . Just like the weights, thethreshold is a real number which is a parameter of the neuron. To put it in more precise algebraic terms:That's all there is to how a perceptron works!That's the basic mathematical model. A way you can think about the perceptron is that it's a device that makes decisions by weighing up evidence. Let me give an example. It's not a veryrealistic example, but it's easy to understand, and we'll soon get to more realistic examples. Suppose the weekend is coming up, and you've heard that there's going to be a cheese festival in your city.You like cheese, and are trying to decide whether or not to go to the festival. You might make your decision by weighing up three factors:1. Is the weather good?2. Does your boyfriend or girlfriend want to accompany you?3. Is the festival near public transit? (You don't own a car).We can represent these three factors by corresponding binary variables , and . For instance, we'd have if the weather is good, and if the weather is bad. Similarly, if your boyfriend or girlfriend wants to go, and if not. And similarly again for and public transit.Now, suppose you absolutely adore cheese, so much so that you're happy to go to the festival even if your boyfriend or girlfriend is uninterested and the festival is hard to get to. But perhaps you really loathe bad weather, and there's no way you'd go to the festival if the weather is bad. You can use perceptrons to model this kind of decision-making. One way to do this is to choose a weight for the weather, and and for the other conditions. The larger value of indicates that the weather matters a lot to you, much more than whether your boyfriend or girlfriend joins you, or the nearness of public transit. Finally,suppose you choose a threshold of for the perceptron. With these∑j w j x j output =⎧⎩⎨⎪⎪⎪⎪⎪⎪01if ≤ threshold ∑j w j x j if > threshold∑j w j x j (1),x 1x 2x 3=1x 1=0x 1=1x 2=0x 2x 3=6w 1=2w 2=2w 3w 15choices, the perceptron implements the desired decision-making model, outputting whenever the weather is good, and whenever the weather is bad. It makes no difference to the output whether your boyfriend or girlfriend wants to go, or whether public transit is nearby.By varying the weights and the threshold, we can get different models of decision-making. For example, suppose we insteadchose a threshold of . Then the perceptron would decide that you should go to the festival whenever the weather was good or when both the festival was near public transit and your boyfriend or girlfriend was willing to join you. In other words, it'd be a different model of decision-making. Dropping the threshold means you're more willing to go to the festival.Obviously, the perceptron isn't a complete model of human decision-making! But what the example illustrates is how a perceptron can weigh up different kinds of evidence in order to make decisions. And it should seem plausible that a complexnetwork of perceptrons could make quite subtle decisions:In this network, the first column of perceptrons - what we'll call the first layer of perceptrons - is making three very simple decisions, by weighing the input evidence. What about the perceptrons in the second layer? Each of those perceptrons ismaking a decision by weighing up the results from the first layer of decision-making. In this way a perceptron in the second layer can make a decision at a more complex and more abstract level than perceptrons in the first layer. And even more complex decisions can be made by the perceptron in the third layer. In this way, a many-layer network of perceptrons can engage in sophisticated decision making.Incidentally, when I defined perceptrons I said that a perceptron has just a single output. In the network above the perceptrons look 103like they have multiple outputs. In fact, they're still single output.The multiple output arrows are merely a useful way of indicating that the output from a perceptron is being used as the input to several other perceptrons. It's less unwieldy than drawing a single output line which then splits.Let's simplify the way we describe perceptrons. The condition is cumbersome, and we can make two notational changes to simplify it. The first change is to write as a dot product, , where and are vectors whose components are the weights and inputs,respectively. The second change is to move the threshold to the other side of the inequality, and to replace it by what's known as the perceptron's bias , . Using the bias instead of the threshold, the perceptron rule can be rewritten:You can think of the bias as a measure of how easy it is to get the perceptron to output a . Or to put it in more biological terms, the bias is a measure of how easy it is to get the perceptron to fire . For a perceptron with a really big bias, it's extremely easy for the perceptron to output a . But if the bias is very negative, then it's difficult for the perceptron to output a . Obviously, introducing the bias is only a small change in how we describe perceptrons, but we'll see later that it leads to further notational simplifications.Because of this, in the remainder of the book we won't use the threshold, we'll always use the bias.I've described perceptrons as a method for weighing evidence to make decisions. Another way perceptrons can be used is tocompute the elementary logical functions we usually think of as underlying computation, functions such as AND , OR , and NAND .For example, suppose we have a perceptron with two inputs, each with weight , and an overall bias of. Here's our perceptron:Then we see that input produces output , sinceis positive. Here, I've introduced the >threshold ∑j w j x j ∑j w j x j w ⋅x ≡∑j w j x j w x b ≡−threshold output ={01if w ⋅x +b ≤0if w ⋅x +b >0(2)111−23001(−2)∗0+(−2)∗0+3=3∗symbol to make the multiplications explicit. Similar calculations show that the inputs and produce output . But the input produces output , since is negative.And so our perceptron implements a NAND gate!The NAND example shows that we can use perceptrons tocompute simple logical functions. In fact, we can use networks of perceptrons to compute any logical function at all. The reason is that the NAND gate is universal for computation, that is, we can build any computation up out of NAND gates. For example, we can use NAND gates to build a circuit which adds two bits, and .This requires computing the bitwise sum, , as well as a carry bit which is set to when both and are , i.e., the carry bit is just the bitwise product:To get an equivalent network of perceptrons we replace all the NAND gates by perceptrons with two inputs, each with weight ,and an overall bias of . Here's the resulting network. Note that I've moved the perceptron corresponding to the bottom rightNAND gate a little, just to make it easier to draw the arrows on thediagram:One notable aspect of this network of perceptrons is that the output from the leftmost perceptron is used twice as input to the bottommost perceptron. When I defined the perceptron model I didn't say whether this kind of double-output-to-the-same-place was allowed. Actually, it doesn't much matter. If we don't want to allow this kind of thing, then it's possible to simply merge the two lines, into a single connection with a weight of -4 instead of two connections with -2 weights. (If you don't find this obvious, you 01101110(−2)∗1+(−2)∗1+3=−1x 1x 2⊕x 1x 21x 1x 21x 1x 2−23should stop and prove to yourself that this is equivalent.) With that change, the network looks as follows, with all unmarkedweights equal to -2, all biases equal to 3, and a single weight of -4,as marked:Up to now I've been drawing inputs like and as variables floating to the left of the network of perceptrons. In fact, it'sconventional to draw an extra layer of perceptrons - the input layer- to encode the inputs:This notation for input perceptrons, in which we have an output,but no inputs,is a shorthand. It doesn't actually mean a perceptron with no inputs. To see this, suppose we did have a perceptron with no inputs. Then the weighted sum would always be zero, and so the perceptron would output if , and if . That is,the perceptron would simply output a fixed value, not the desired valued (, in the example above). It's better to think of the input perceptrons as not really being perceptrons at all, but rather special units which are simply defined to output the desired values, .The adder example demonstrates how a network of perceptrons can be used to simulate a circuit containing many NAND gates.And because NAND gates are universal for computation, it follows that perceptrons are also universal for computation.The computational universality of perceptrons is simultaneously x 1x 2∑j w j x j 1b >00b ≤0x 1,,…x 1x 2reassuring and disappointing. It's reassuring because it tells us that networks of perceptrons can be as powerful as any other computing device. But it's also disappointing, because it makes it seem as though perceptrons are merely a new type of NAND gate. That's hardly big news!However, the situation is better than this view suggests. It turns out that we can devise learning algorithms which can automatically tune the weights and biases of a network of artificial neurons. This tuning happens in response to external stimuli, without direct intervention by a programmer. These learning algorithms enable us to use artificial neurons in a way which is radically different to conventional logic gates. Instead of explicitly laying out a circuit of NAND and other gates, our neural networks can simply learn to solve problems, sometimes problems where it would be extremely difficult to directly design a conventional circuit.Sigmoid neuronsLearning algorithms sound terrific. But how can we devise such algorithms for a neural network? Suppose we have a network of perceptrons that we'd like to use to learn to solve some problem. For example, the inputs to the network might be the raw pixel data from a scanned, handwritten image of a digit. And we'd like the network to learn weights and biases so that the output from the network correctly classifies the digit. To see how learning might work, suppose we make a small change in some weight (or bias) in the network. What we'd like is for this small change in weight to cause only a small corresponding change in the output from the network. As we'll see in a moment, this property will make learning possible. Schematically, here's what we want (obviously this network is too simple to do handwriting recognition!):If it were true that a small change in a weight (or bias) causes only a small change in output, then we could use this fact to modify the weights and biases to get our network to behave more in the manner we want. For example, suppose the network was mistakenly classifying an image as an "8" when it should be a "9". We could figure out how to make a small change in the weights and biases so the network gets a little closer to classifying the image as a "9". And then we'd repeat this, changing the weights and biases over and over to produce better and better output. The network would be learning.The problem is that this isn't what happens when our network contains perceptrons. In fact, a small change in the weights or bias of any single perceptron in the network can sometimes cause the01 output of that perceptron to completely flip, say from to . That flip may then cause the behaviour of the rest of the network to completely change in some very complicated way. So while your "9" might now be classified correctly, the behaviour of the network on all the other images is likely to have completely changed in some hard-to-control way. That makes it difficult to see how to gradually modify the weights and biases so that the network gets closer to the desired behaviour. Perhaps there's some clever way of getting around this problem. But it's not immediately obvious how we can get a network of perceptrons to learn.We can overcome this problem by introducing a new type of artificial neuron called a sigmoid neuron. Sigmoid neurons are similar to perceptrons, but modified so that small changes in their weights and bias cause only a small change in their output. That's the crucial fact which will allow a network of sigmoid neurons to learn.Okay, let me describe the sigmoid neuron. We'll depict sigmoid neurons in the same way we depicted perceptrons:Just like a perceptron, the sigmoid neuron has inputs, .But instead of being just or , these inputs can also take on any values between and . So, for instance, is a valid input for a sigmoid neuron. Also just like a perceptron, the sigmoid neuron has weights for each input, , and an overall bias, . But the output is not or . Instead, it's , where is called the sigmoid function *, and is defined by:To put it all a little more explicitly, the output of a sigmoid neuron with inputs , weights , and bias isAt first sight, sigmoid neurons appear very different toperceptrons. The algebraic form of the sigmoid function may seem opaque and forbidding if you're not already familiar with it. In fact,there are many similarities between perceptrons and sigmoidneurons, and the algebraic form of the sigmoid function turns out to be more of a technical detail than a true barrier tounderstanding.To understand the similarity to the perceptron model, suppose is a large positive number. Then and so . In other words, when is large and positive,the output from the sigmoid neuron is approximately , just as it would have been for a perceptron. Suppose on the other hand that is very negative. Then , and . So when is very negative, the behaviour of a sigmoid neuron also closely approximates a perceptron. It's only when is of modest size that there's much deviation from the perceptron model.,,…x 1x 201010.638…,,…w 1w 2b 01σ(w ⋅x +b )σ*Incidentally , is sometimes called the logistic function , and this new class of neurons called logistic neurons . It's useful to remember this terminology , since these terms are used by many people working with neural nets. Howev er, we'll stick with the sigmoid terminology.σσ(z )≡.11+e −z (3),,…x 1x 2,,…w 1w 2b .11+exp(−−b )∑j w j x j (4)z ≡w ⋅x +b ≈0e −z σ(z )≈1z =w ⋅x +b 1z =w ⋅x +b →∞e −z σ(z )≈0z =w ⋅x +b w ⋅x +bwith all the partial derivatives, it's actually saying something very simple (and which is very good news): is a linear function of the changes and in the weights and bias. This linearity makes it easy to choose small changes in the weights and biases to achieve any desired small change in the output. So while sigmoid neurons have much of the same qualitative behaviour as perceptrons, they make it much easier to figure out how changing the weights and biases will change the output.If it's the shape of which really matters, and not its exact form,then why use the particular form used for in Equation (3)? In fact, later in the book we will occasionally consider neurons where the output is for some other activation function .The main thing that changes when we use a different activation function is that the particular values for the partial derivatives in Equation (5) change. It turns out that when we compute those partial derivatives later, using will simplify the algebra, simply because exponentials have lovely properties when differentiated.In any case, is commonly-used in work on neural nets, and is the activation function we'll use most often in this book.How should we interpret the output from a sigmoid neuron?Obviously, one big difference between perceptrons and sigmoid neurons is that sigmoid neurons don't just output or . They can have as output any real number between and , so values such as and are legitimate outputs. This can be useful,for example, if we want to use the output value to represent the average intensity of the pixels in an image input to a neuralnetwork. But sometimes it can be a nuisance. Suppose we want the output from the network to indicate either "the input image is a 9"or "the input image is not a 9". Obviously, it'd be easiest to do this if the output was a or a , as in a perceptron. But in practice we can set up a convention to deal with this, for example, by deciding to interpret any output of at least as indicating a "9", and any output less than as indicating "not a 9". I'll always explicitly state when we're using such a convention, so it shouldn't cause any confusion.ExercisesSigmoid neurons simulating perceptrons, part IΔoutput Δw j Δb σσf (w ⋅x +b )f (⋅)σσ01010.173…0.689…010.50.5Suppose we take all the weights and biases in a network of perceptrons, and multiply them by a positive constant, .Show that the behaviour of the network doesn't change.Sigmoid neurons simulating perceptrons, part II Suppose we have the same setup as the last problem - anetwork of perceptrons. Suppose also that the overall input to the network of perceptrons has been chosen. We won't need the actual input value, we just need the input to have been fixed. Suppose the weights and biases are such thatfor the input to any particular perceptron in the network. Now replace all the perceptrons in the network by sigmoid neurons, and multiply the weights and biases by a positive constant . Show that in the limit as the behaviour of this network of sigmoid neurons is exactly the same as the network of perceptrons. How can this fail when for one of the perceptrons?The architecture of neural networksIn the next section I'll introduce a neural network that can do a pretty good job classifying handwritten digits. In preparation for that, it helps to explain some terminology that lets us name different parts of a network. Suppose we have the network:As mentioned earlier, the leftmost layer in this network is called the input layer, and the neurons within the layer are called input neurons . The rightmost or output layer contains the output neurons , or, as in this case, a single output neuron. The middle layer is called a hidden layer , since the neurons in this layer are neither inputs nor outputs. The term "hidden" perhaps sounds a little mysterious - the first time I heard the term I thought it must have some deep philosophical or mathematical significance - but it really means nothing more than "not an input or an output". The c >0w ⋅x +b ≠0x c >0c →∞w ⋅x +b =0network above has just a single hidden layer, but some networks have multiple hidden layers. For example, the following four-layernetwork has two hidden layers:Somewhat confusingly, and for historical reasons, such multiple layer networks are sometimes called multilayer perceptrons or MLPs , despite being made up of sigmoid neurons, not perceptrons.I'm not going to use the MLP terminology in this book, since I think it's confusing, but wanted to warn you of its existence.The design of the input and output layers in a network is often straightforward. For example, suppose we're trying to determine whether a handwritten image depicts a "9" or not. A natural way to design the network is to encode the intensities of the image pixels into the input neurons. If the image is a by greyscale image,then we'd have input neurons, with theintensities scaled appropriately between and . The output layer will contain just a single neuron, with output values of less than indicating "input image is not a 9", and values greater than indicating "input image is a 9 ".While the design of the input and output layers of a neuralnetwork is often straightforward, there can be quite an art to the design of the hidden layers. In particular, it's not possible to sum up the design process for the hidden layers with a few simple rules of thumb. Instead, neural networks researchers have developed many design heuristics for the hidden layers, which help people get the behaviour they want out of their nets. For example, such heuristics can be used to help determine how to trade off the number of hidden layers against the time required to train the network. We'll meet several such design heuristics later in this book.64644,096=64×64010.50.5Up to now, we've been discussing neural networks where the output from one layer is used as input to the next layer. Such networks are called feedforward neural networks. This means there are no loops in the network - information is always fed forward, never fed back. If we did have loops, we'd end up withσsituations where the input to the function depended on the output. That'd be hard to make sense of, and so we don't allow such loops.However, there are other models of artificial neural networks in which feedback loops are possible. These models are called recurrent neural networks. The idea in these models is to have neurons which fire for some limited duration of time, before becoming quiescent. That firing can stimulate other neurons, which may fire a little while later, also for a limited duration. That causes still more neurons to fire, and so over time we get a cascade of neurons firing. Loops don't cause problems in such a model, since a neuron's output only affects its input at some later time, not instantaneously.Recurrent neural nets have been less influential than feedforward networks, in part because the learning algorithms for recurrent nets are (at least to date) less powerful. But recurrent networks are still extremely interesting. They're much closer in spirit to how our brains work than feedforward networks. And it's possible that recurrent networks can solve important problems which can only be solved with great difficulty by feedforward networks. However, to limit our scope, in this book we're going to concentrate on the more widely-used feedforward networks.A simple network to classify handwritten digitsHaving defined neural networks, let's return to handwriting recognition. We can split the problem of recognizing handwritten digits into two sub-problems. First, we'd like a way of breaking an image containing many digits into a sequence of separate images, each containing a single digit. For example, we'd like to break the image。
gradient-based learning applied to document recognition
Yann LeCun, Leon Bottou, Yoshua Bengio, and Patrick Ha nral Networks trained with the backpropagation algorithm constitute the best example of a successful Gradient-Based Learning technique. Given an appropriate network architecture, Gradient-Based Learning algorithms can be used to synthesize a complex decision surface that can classify high-dimensional patterns such as handwritten characters, with minimal preprocessing. This paper reviews various methods applied to handwritten character recognition and compares them on a standard handwritten digit recognition task. Convolutional Neural Networks, that are speci cally designed to deal with the variability of 2D shapes, are shown to outperform all other techniques. Real-life document recognition systems are composed of multiple modules including eld extraction, segmentation, recognition, and language modeling. A new learning paradigm, called Graph Transformer Networks (GTN), allows such multi-module systems to be trained globally using GradientBased methods so as to minimize an overall performance measure. Two systems for on-line handwriting recognition are described. Experiments demonstrate the advantage of global training, and the exibility of Graph Transformer Networks. A Graph Transformer Network for reading bank check is also described. It uses Convolutional Neural Network character recognizers combined with global training techniques to provides record accuracy on business and personal checks. It is deployed commercially and reads million of checks per month.
英语四级仔细阅读方法和技巧
• had to be answered with any commercial consequence was when the laying of a telegraph cable from Europe to America was proposed. The engineers had to know the depth profile( 轮廓)of the route to estimate the length of cable that had to be manufactured. It was to Maury of US Navy that the Atlantic Telegraph Company turned, in 1853 , for information on this matter. In the 1840s,Maury had been responsible for encouraging voyages during which soundings(试探)were taken investigate the depths of the North Atlantic and Pacific Oceans. Later, some of his findings aroused much popular interest in his book The Physical Geography of the Sea. The cable was laid , but not until 1866 was the connection made permanent and reliable. At the early attempts, the cable failed and when it was taken out for repairs it was found to be covered in living growths, a fact which defied contemporary scientific opinion that there was no life in the deeper parts of the sea.
文献翻译四One Maya. More Value.
<文献翻译：原文>One Maya. More V alue.Autodesk Maya 2010 software is the first release to unify the Maya Complete 2009 and Maya Unlimited 2009 feature sets, advanced matchmoving capabilities , and powerful compositing into a single offering with exceptional value. Producers have become more savvy with respect to computer-generated imagery; they expect more work with additional complexity in less time than ever before. Maya 2010 gives us the total package to efficiently handle any challenge they can throw at us, whether it’s heavy in tracking, modeling, animating, rendering, or compositing.—Paal AnandDigital Post SupervisorBling Imaging(人名和公司名不用翻译) For those looking to create the kind of compelling digital imagery found in Academy Award®−winning films and top-selling games, Autodesk® Maya® 2010 delivers extensive 3D modeling, animation, and rendering toolsets; innovative simulation and compositing technologies; and a flexible software development kit (SDK) and scripting capabilities —making it easier, and more affordable, to create stylistic designs, believable animated characters, and lifelike visual effects.Unbeatable V alueAutodesk Maya 2010 offers an end-to-end computer graphics (CG) workflow based on the award-winning Maya Unlimited 2009 toolset, with its advanced simulation tools for cloth, hair, fur, fluids, and particles. To supplement your creative workflow, we’ve also added t he Maya Composite high dynamic range compositing system, a matchmoving camera tracking system, five additional mental ray® for Maya batch* rendering nodes, and the Autodesk® Backburner™** network render queue manager. Proven SolutionMaya has been a favorite among companies producing top film, games, and television content throughout the world for the last decade Meanwhile, award-winning commercial spot facilities like The Mill and Psyop count Maya among their toolsets, as do top broadcasters such as NBC, Seven Network, and Turner.FeaturesMaya has been at the cutting edge of feature development for over 10 years, and Maya 2010 is no exception. The software is packed with tried and tested features that help speed your project from initial concept to finished renderings: polygon andNURBS modeling, UV mapping and texturing, animation and rigging, dynamic simulation tools, tools for generating plants and other natural detail, in addition to advanced compositing capabilities, and a choice of four built-in renderers, including mental ray.ProductivityIncreased competition for projects and tighter deadlines mean that many jobs require even more high-quality work in less time. Maya 2010 helps maximize productivity with optimized workflows for everyday tasks; opportunities for collaborative, parallel workflows and reuse of assets; and automation of repetitive tasks through scripting. PerformanceThrough a combination of multi-threading, algorithmic tuning, sophisticated memory management, and tools for segmenting scenes, Maya 2010 is engineered to elegantly handle today’s increasingly complex data sets without restricting the creative process. InteroperabilityWhether you are painting textures in Adobe® Photoshop® software, compositing shots in Maya Composite or Autodesk® Flame® software, or bringing in cleaned motion capture data from Autodesk® MotionBuilder® character animation software, Maya 2010 helps to minimize errors and reduce iterations. And, support for the Autodesk® FBX® data interchange technology enables you to reuse assets created outside of Maya in your Maya scenes. Maya also offers an SDK to assist with pipeline integration.ExtensibilityMaya is an out-of-the-box solution, but for companies that want to integrate it with their pipelines, or to add new features, Maya offers avenues for customization. Built from the ground up with its own embedded scripting language, Maya Embedded Language (MEL), Maya 2010 also offers Python® scripting and an extensive, well-documented C++ application programming interface (API).Platform of ChoiceWhether you use a Windows®, Mac®, or Linux® operating system, Maya 2010 runs on your platform of choice. And it’s ready to handle the large amounts of memory that today’s large scenes require, with64-bit executables for both Windows and Linuxoperating systems.Advanced Simulation ToolsEvery license of Maya 2010 now includes the innovative Maya Nucleus unified simulation framework and the first two fully integrated Nucleus modules—Maya nCloth and Maya nParticles—as well as Maya Fluid Effects, Maya Hair, and Maya Fur. These widely used, production-proven toolsets for simulating cloth, fluids, hair, and fur enable you to more efficiently create the types of sophisticated effects audiences crave, without additional software investment.High-Performance CompositingMaya Composite brings high-performance, high dynamic range (HDR) compositing to Maya 2010. The comprehensive Maya Composite toolset gives you keying, tracking, color correction, rotoscoping, paint, and warping tools; advanced filters (including motion blur and depth of field); a full 3D compositing environment; and support for stereoscopic production. Available on the same choice of platforms as Maya, this node-based compositor provides you with a high-efficiency, collaborative compositing environment.Professional Camera TrackingA crucial tool for any leading visual effects production work, Autodesk® MatchMover™software makes high-quality 3D camera tracking accessible within Maya. Using this toolset, you can extract accurate 3D camera and motion data from video and film sequences so you can insert your Maya elements seamlessly into the footage. MatchMover combines automatic tracking capabilities with the precision manual controls professionals demand.Augmented Rendering PowerWith five additional mental ray for Maya batch rendering nodes, you can now use a network of computers to render sequences faster. The Backburner network render queue manager is also included with Maya 2010, to help you manage the process; or simply integrate the additional mental ray for Maya nodes with your existing render management software.<文献翻译：译文>同一个Maya，更多的价值Autodesk Maya 2010软件是首次发布的统一了Maya Complete 2009和Maya Unlimited 2009特性的集合，先进的镜头跟踪能力和强有力的合成一个单一的具有额外价值的供给能力。
Gradient-based learning applied to document recognition
Gradient-Based Learning Appliedto Document RecognitionYANN LECUN,MEMBER,IEEE,L´EON BOTTOU,YOSHUA BENGIO,AND PATRICK HAFFNER Invited PaperMultilayer neural networks trained with the back-propagation algorithm constitute the best example of a successful gradient-based learning technique.Given an appropriate network architecture,gradient-based learning algorithms can be used to synthesize a complex decision surface that can classify high-dimensional patterns,such as handwritten characters,with minimal preprocessing.This paper reviews various methods applied to handwritten character recognition and compares them on a standard handwritten digit recognition task.Convolutional neural networks,which are specifically designed to deal with the variability of two dimensional(2-D)shapes,are shown to outperform all other techniques.Real-life document recognition systems are composed of multiple modules includingfield extraction,segmentation,recognition, and language modeling.A new learning paradigm,called graph transformer networks(GTN’s),allows such multimodule systems to be trained globally using gradient-based methods so as to minimize an overall performance measure.Two systems for online handwriting recognition are described. Experiments demonstrate the advantage of global training,and theflexibility of graph transformer networks.A graph transformer network for reading a bank check is also described.It uses convolutional neural network character recognizers combined with global training techniques to provide record accuracy on business and personal checks.It is deployed commercially and reads several million checks per day. Keywords—Convolutional neural networks,document recog-nition,finite state transducers,gradient-based learning,graphtransformer networks,machine learning,neural networks,optical character recognition(OCR).N OMENCLATUREGT Graph transformer.GTN Graph transformer network.HMM Hidden Markov model.HOS Heuristic oversegmentation.K-NN K-nearest neighbor.Manuscript received November1,1997;revised April17,1998.Y.LeCun,L.Bottou,and P.Haffner are with the Speech and Image Processing Services Research Laboratory,AT&T Labs-Research,Red Bank,NJ07701USA.Y.Bengio is with the D´e partement d’Informatique et de Recherche Op´e rationelle,Universit´e de Montr´e al,Montr´e al,Qu´e bec H3C3J7Canada. Publisher Item Identifier S0018-9219(98)07863-3.NN Neural network.OCR Optical character recognition.PCA Principal component analysis.RBF Radial basis function.RS-SVM Reduced-set support vector method. SDNN Space displacement neural network.SVM Support vector method.TDNN Time delay neural network.V-SVM Virtual support vector method.I.I NTRODUCTIONOver the last several years,machine learning techniques, particularly when applied to NN’s,have played an increas-ingly important role in the design of pattern recognition systems.In fact,it could be argued that the availability of learning techniques has been a crucial factor in the recent success of pattern recognition applications such as continuous speech recognition and handwriting recognition. The main message of this paper is that better pattern recognition systems can be built by relying more on auto-matic learning and less on hand-designed heuristics.This is made possible by recent progress in machine learning and computer ing character recognition as a case study,we show that hand-crafted feature extraction can be advantageously replaced by carefully designed learning machines that operate directly on pixel ing document understanding as a case study,we show that the traditional way of building recognition systems by manually integrating individually designed modules can be replaced by a unified and well-principled design paradigm,called GTN’s,which allows training all the modules to optimize a global performance criterion.Since the early days of pattern recognition it has been known that the variability and richness of natural data, be it speech,glyphs,or other types of patterns,make it almost impossible to build an accurate recognition system entirely by hand.Consequently,most pattern recognition systems are built using a combination of automatic learning techniques and hand-crafted algorithms.The usual method0018–9219/98$10.00©1998IEEE2278PROCEEDINGS OF THE IEEE,VOL.86,NO.11,NOVEMBER1998Fig.1.Traditional pattern recognition is performed with two modules:afixed feature extractor and a trainable classifier.of recognizing individual patterns consists in dividing the system into two main modules shown in Fig.1.Thefirst module,called the feature extractor,transforms the input patterns so that they can be represented by low-dimensional vectors or short strings of symbols that:1)can be easily matched or compared and2)are relatively invariant with respect to transformations and distortions of the input pat-terns that do not change their nature.The feature extractor contains most of the prior knowledge and is rather specific to the task.It is also the focus of most of the design effort, because it is often entirely hand crafted.The classifier, on the other hand,is often general purpose and trainable. One of the main problems with this approach is that the recognition accuracy is largely determined by the ability of the designer to come up with an appropriate set of features. This turns out to be a daunting task which,unfortunately, must be redone for each new problem.A large amount of the pattern recognition literature is devoted to describing and comparing the relative merits of different feature sets for particular tasks.Historically,the need for appropriate feature extractors was due to the fact that the learning techniques used by the classifiers were limited to low-dimensional spaces with easily separable classes[1].A combination of three factors has changed this vision over the last decade.First, the availability of low-cost machines with fast arithmetic units allows for reliance on more brute-force“numerical”methods than on algorithmic refinements.Second,the avail-ability of large databases for problems with a large market and wide interest,such as handwriting recognition,has enabled designers to rely more on real data and less on hand-crafted feature extraction to build recognition systems. The third and very important factor is the availability of powerful machine learning techniques that can handle high-dimensional inputs and can generate intricate decision functions when fed with these large data sets.It can be argued that the recent progress in the accuracy of speech and handwriting recognition systems can be attributed in large part to an increased reliance on learning techniques and large training data sets.As evidence of this fact,a large proportion of modern commercial OCR systems use some form of multilayer NN trained with back propagation.In this study,we consider the tasks of handwritten character recognition(Sections I and II)and compare the performance of several learning techniques on a benchmark data set for handwritten digit recognition(Section III). While more automatic learning is beneficial,no learning technique can succeed without a minimal amount of prior knowledge about the task.In the case of multilayer NN’s, a good way to incorporate knowledge is to tailor its archi-tecture to the task.Convolutional NN’s[2],introduced in Section II,are an example of specialized NN architectures which incorporate knowledge about the invariances of two-dimensional(2-D)shapes by using local connection patterns and by imposing constraints on the weights.A comparison of several methods for isolated handwritten digit recogni-tion is presented in Section III.To go from the recognition of individual characters to the recognition of words and sentences in documents,the idea of combining multiple modules trained to reduce the overall error is introduced in Section IV.Recognizing variable-length objects such as handwritten words using multimodule systems is best done if the modules manipulate directed graphs.This leads to the concept of trainable GTN,also introduced in Section IV. Section V describes the now classical method of HOS for recognizing words or other character strings.Discriminative and nondiscriminative gradient-based techniques for train-ing a recognizer at the word level without requiring manual segmentation and labeling are presented in Section VI. Section VII presents the promising space-displacement NN approach that eliminates the need for segmentation heuris-tics by scanning a recognizer at all possible locations on the input.In Section VIII,it is shown that trainable GTN’s can be formulated as multiple generalized transductions based on a general graph composition algorithm.The connections between GTN’s and HMM’s,commonly used in speech recognition,is also treated.Section IX describes a globally trained GTN system for recognizing handwriting entered in a pen computer.This problem is known as “online”handwriting recognition since the machine must produce immediate feedback as the user writes.The core of the system is a convolutional NN.The results clearly demonstrate the advantages of training a recognizer at the word level,rather than training it on presegmented, hand-labeled,isolated characters.Section X describes a complete GTN-based system for reading handwritten and machine-printed bank checks.The core of the system is the convolutional NN called LeNet-5,which is described in Section II.This system is in commercial use in the NCR Corporation line of check recognition systems for the banking industry.It is reading millions of checks per month in several banks across the United States.A.Learning from DataThere are several approaches to automatic machine learn-ing,but one of the most successful approaches,popularized in recent years by the NN community,can be called“nu-merical”or gradient-based learning.The learning machine computes afunction th input pattern,andtheoutputthatminimizesand the error rate on the trainingset decreases with the number of training samplesapproximatelyasis the number of trainingsamples,is a number between0.5and1.0,andincreases,decreases.Therefore,when increasing thecapacitythat achieves the lowest generalizationerror Mostlearning algorithms attempt tominimize as well assome estimate of the gap.A formal version of this is calledstructural risk minimization[6],[7],and it is based on defin-ing a sequence of learning machines of increasing capacity,corresponding to a sequence of subsets of the parameterspace such that each subset is a superset of the previoussubset.In practical terms,structural risk minimization isimplemented byminimizingisaconstant.that belong to high-capacity subsets ofthe parameter space.Minimizingis a real-valuedvector,with respect towhichis iteratively adjusted asfollows:is updated on the basis of a singlesampleof several layers of processing,i.e.,the back-propagation algorithm.The third event was the demonstration that the back-propagation procedure applied to multilayer NN’s with sigmoidal units can solve complicated learning tasks. The basic idea of back propagation is that gradients can be computed efficiently by propagation from the output to the input.This idea was described in the control theory literature of the early1960’s[16],but its application to ma-chine learning was not generally realized then.Interestingly, the early derivations of back propagation in the context of NN learning did not use gradients but“virtual targets”for units in intermediate layers[17],[18],or minimal disturbance arguments[19].The Lagrange formalism used in the control theory literature provides perhaps the best rigorous method for deriving back propagation[20]and for deriving generalizations of back propagation to recurrent networks[21]and networks of heterogeneous modules[22].A simple derivation for generic multilayer systems is given in Section I-E.The fact that local minima do not seem to be a problem for multilayer NN’s is somewhat of a theoretical mystery. It is conjectured that if the network is oversized for the task(as is usually the case in practice),the presence of “extra dimensions”in parameter space reduces the risk of unattainable regions.Back propagation is by far the most widely used neural-network learning algorithm,and probably the most widely used learning algorithm of any form.D.Learning in Real Handwriting Recognition Systems Isolated handwritten character recognition has been ex-tensively studied in the literature(see[23]and[24]for reviews),and it was one of the early successful applications of NN’s[25].Comparative experiments on recognition of individual handwritten digits are reported in Section III. They show that NN’s trained with gradient-based learning perform better than all other methods tested here on the same data.The best NN’s,called convolutional networks, are designed to learn to extract relevant features directly from pixel images(see Section II).One of the most difficult problems in handwriting recog-nition,however,is not only to recognize individual charac-ters,but also to separate out characters from their neighbors within the word or sentence,a process known as seg-mentation.The technique for doing this that has become the“standard”is called HOS.It consists of generating a large number of potential cuts between characters using heuristic image processing techniques,and subsequently selecting the best combination of cuts based on scores given for each candidate character by the recognizer.In such a model,the accuracy of the system depends upon the quality of the cuts generated by the heuristics,and on the ability of the recognizer to distinguish correctly segmented characters from pieces of characters,multiple characters, or otherwise incorrectly segmented characters.Training a recognizer to perform this task poses a major challenge because of the difficulty in creating a labeled database of incorrectly segmented characters.The simplest solution consists of running the images of character strings through the segmenter and then manually labeling all the character hypotheses.Unfortunately,not only is this an extremely tedious and costly task,it is also difficult to do the labeling consistently.For example,should the right half of a cut-up four be labeled as a one or as a noncharacter?Should the right half of a cut-up eight be labeled as a three?Thefirst solution,described in Section V,consists of training the system at the level of whole strings of char-acters rather than at the character level.The notion of gradient-based learning can be used for this purpose.The system is trained to minimize an overall loss function which measures the probability of an erroneous answer.Section V explores various ways to ensure that the loss function is differentiable and therefore lends itself to the use of gradient-based learning methods.Section V introduces the use of directed acyclic graphs whose arcs carry numerical information as a way to represent the alternative hypotheses and introduces the idea of GTN.The second solution,described in Section VII,is to eliminate segmentation altogether.The idea is to sweep the recognizer over every possible location on the input image,and to rely on the“character spotting”property of the recognizer,i.e.,its ability to correctly recognize a well-centered character in its inputfield,even in the presence of other characters besides it,while rejecting images containing no centered characters[26],[27].The sequence of recognizer outputs obtained by sweeping the recognizer over the input is then fed to a GTN that takes linguistic constraints into account andfinally extracts the most likely interpretation.This GTN is somewhat similar to HMM’s,which makes the approach reminiscent of the classical speech recognition[28],[29].While this technique would be quite expensive in the general case,the use of convolutional NN’s makes it particularly attractive because it allows significant savings in computational cost.E.Globally Trainable SystemsAs stated earlier,most practical pattern recognition sys-tems are composed of multiple modules.For example,a document recognition system is composed of afield loca-tor(which extracts regions of interest),afield segmenter (which cuts the input image into images of candidate characters),a recognizer(which classifies and scores each candidate character),and a contextual postprocessor,gen-erally based on a stochastic grammar(which selects the best grammatically correct answer from the hypotheses generated by the recognizer).In most cases,the information carried from module to module is best represented as graphs with numerical information attached to the arcs. For example,the output of the recognizer module can be represented as an acyclic graph where each arc contains the label and the score of a candidate character,and where each path represents an alternative interpretation of the input string.Typically,each module is manually optimized,or sometimes trained,outside of its context.For example,the character recognizer would be trained on labeled images of presegmented characters.Then the complete system isLECUN et al.:GRADIENT-BASED LEARNING APPLIED TO DOCUMENT RECOGNITION2281assembled,and a subset of the parameters of the modules is manually adjusted to maximize the overall performance. This last step is extremely tedious,time consuming,and almost certainly suboptimal.A better alternative would be to somehow train the entire system so as to minimize a global error measure such as the probability of character misclassifications at the document level.Ideally,we would want tofind a good minimum of this global loss function with respect to all theparameters in the system.If the loss functionusing gradient-based learning.However,at first glance,it appears that the sheer size and complexity of the system would make this intractable.To ensure that the global loss functionwithrespect towith respect toFig.2.Architecture of LeNet-5,a convolutional NN,here used for digits recognition.Each plane is a feature map,i.e.,a set of units whose weights are constrained to be identical.or other2-D or one-dimensional(1-D)signals,must be approximately size normalized and centered in the input field.Unfortunately,no such preprocessing can be perfect: handwriting is often normalized at the word level,which can cause size,slant,and position variations for individual characters.This,combined with variability in writing style, will cause variations in the position of distinctive features in input objects.In principle,a fully connected network of sufficient size could learn to produce outputs that are invari-ant with respect to such variations.However,learning such a task would probably result in multiple units with similar weight patterns positioned at various locations in the input so as to detect distinctive features wherever they appear on the input.Learning these weight configurations requires a very large number of training instances to cover the space of possible variations.In convolutional networks,as described below,shift invariance is automatically obtained by forcing the replication of weight configurations across space. Secondly,a deficiency of fully connected architectures is that the topology of the input is entirely ignored.The input variables can be presented in any(fixed)order without af-fecting the outcome of the training.On the contrary,images (or time-frequency representations of speech)have a strong 2-D local structure:variables(or pixels)that are spatially or temporally nearby are highly correlated.Local correlations are the reasons for the well-known advantages of extracting and combining local features before recognizing spatial or temporal objects,because configurations of neighboring variables can be classified into a small number of categories (e.g.,edges,corners,etc.).Convolutional networks force the extraction of local features by restricting the receptive fields of hidden units to be local.A.Convolutional NetworksConvolutional networks combine three architectural ideas to ensure some degree of shift,scale,and distortion in-variance:1)local receptivefields;2)shared weights(or weight replication);and3)spatial or temporal subsampling.A typical convolutional network for recognizing characters, dubbed LeNet-5,is shown in Fig.2.The input plane receives images of characters that are approximately size normalized and centered.Each unit in a layer receives inputs from a set of units located in a small neighborhood in the previous layer.The idea of connecting units to local receptivefields on the input goes back to the perceptron in the early1960’s,and it was almost simultaneous with Hubel and Wiesel’s discovery of locally sensitive,orientation-selective neurons in the cat’s visual system[30].Local connections have been used many times in neural models of visual learning[2],[18],[31]–[34].With local receptive fields neurons can extract elementary visual features such as oriented edges,endpoints,corners(or similar features in other signals such as speech spectrograms).These features are then combined by the subsequent layers in order to detect higher order features.As stated earlier,distortions or shifts of the input can cause the position of salient features to vary.In addition,elementary feature detectors that are useful on one part of the image are likely to be useful across the entire image.This knowledge can be applied by forcing a set of units,whose receptivefields are located at different places on the image,to have identical weight vectors[15], [32],[34].Units in a layer are organized in planes within which all the units share the same set of weights.The set of outputs of the units in such a plane is called a feature map. Units in a feature map are all constrained to perform the same operation on different parts of the image.A complete convolutional layer is composed of several feature maps (with different weight vectors),so that multiple features can be extracted at each location.A concrete example of this is thefirst layer of LeNet-5shown in Fig.2.Units in thefirst hidden layer of LeNet-5are organized in six planes,each of which is a feature map.A unit in a feature map has25inputs connected to a5case of LeNet-5,at each input location six different types of features are extracted by six units in identical locations in the six feature maps.A sequential implementation of a feature map would scan the input image with a single unit that has a local receptive field and store the states of this unit at corresponding locations in the feature map.This operation is equivalent to a convolution,followed by an additive bias and squashing function,hence the name convolutional network.The kernel of the convolution is theOnce a feature has been detected,its exact location becomes less important.Only its approximate position relative to other features is relevant.For example,once we know that the input image contains the endpoint of a roughly horizontal segment in the upper left area,a corner in the upper right area,and the endpoint of a roughly vertical segment in the lower portion of the image,we can tell the input image is a seven.Not only is the precise position of each of those features irrelevant for identifying the pattern,it is potentially harmful because the positions are likely to vary for different instances of the character.A simple way to reduce the precision with which the position of distinctive features are encoded in a feature map is to reduce the spatial resolution of the feature map.This can be achieved with a so-called subsampling layer,which performs a local averaging and a subsampling,thereby reducing the resolution of the feature map and reducing the sensitivity of the output to shifts and distortions.The second hidden layer of LeNet-5is a subsampling layer.This layer comprises six feature maps,one for each feature map in the previous layer.The receptive field of each unit is a 232p i x e l i m a g e .T h i s i s s i g n i fic a n tt h e l a r g e s t c h a r a c t e r i n t h e d a t a b a s e (a t28fie l d ).T h e r e a s o n i s t h a t i t it h a t p o t e n t i a l d i s t i n c t i v e f e a t u r e s s u c h o r c o r n e r c a n a p p e a r i n t h e c e n t e r o f t h o f t h e h i g h e s t l e v e l f e a t u r e d e t e c t o r s .o f c e n t e r s o f t h e r e c e p t i v e fie l d s o f t h e l a y e r (C 3,s e e b e l o w )f o r m a 2032i n p u t .T h e v a l u e s o f t h e i n p u t p i x e l s o t h a t t h e b a c k g r o u n d l e v e l (w h i t e )c o ro fa n d t h e f o r e g r o u n d (b l ac k )c o r r e s p T h i s m a k e s t h e m e a n i n p u t r o u g h l y z e r o r o u g h l y o n e ,w h i c h a c c e l e r a t e s l e a r n i n g I n t h e f o l l o w i n g ,c o n v o l u t i o n a l l a y e r s u b s a m p l i n g l a y e r s a r e l a b e l ed S x ,a n d l a ye r s a r e l a b e l e d F x ,w h e r e x i s t h e l a y L a y e r C 1i s a c o n v o l u t i o n a l l a y e r w i t h E a c h u n i t i n e a c hf e a t u r e m a p i s c o n n e c t28w h i c h p r e v e n t s c o n n e c t i o n f r o m t h e i n p t h e b o u n d a r y .C 1c o n t a i n s 156t r a i n a b l 122304c o n n e c t i o n s .L a y e r S 2i s a s u b s a m p l i n g l a y e r w i t h s i s i z e 142n e i g h b o r h o o d i n t h e c o r r e s p o n d i n g f T h e f o u r i n p u t s t o a u n i t i n S 2a r e a d d e d ,2284P R O C E E D I N G S O F T H E I E E E ,V O L .86,N O .11,N O VTable 1Each Column Indicates Which Feature Map in S2Are Combined by the Units in a Particular Feature Map ofC3a trainable coefficient,and then added to a trainable bias.The result is passed through a sigmoidal function.The25neighborhoods at identical locations in a subset of S2’s feature maps.Table 1shows the set of S2feature maps combined by each C3feature map.Why not connect every S2feature map to every C3feature map?The reason is twofold.First,a noncomplete connection scheme keeps the number of connections within reasonable bounds.More importantly,it forces a break of symmetry in the network.Different feature maps are forced to extract dif-ferent (hopefully complementary)features because they get different sets of inputs.The rationale behind the connection scheme in Table 1is the following.The first six C3feature maps take inputs from every contiguous subsets of three feature maps in S2.The next six take input from every contiguous subset of four.The next three take input from some discontinuous subsets of four.Finally,the last one takes input from all S2feature yer C3has 1516trainable parameters and 156000connections.Layer S4is a subsampling layer with 16feature maps of size52neighborhood in the corresponding feature map in C3,in a similar way as C1and yer S4has 32trainable parameters and 2000connections.Layer C5is a convolutional layer with 120feature maps.Each unit is connected to a55,the size of C5’s feature maps is11.This process of dynamically increasing thesize of a convolutional network is described in Section yer C5has 48120trainable connections.Layer F6contains 84units (the reason for this number comes from the design of the output layer,explained below)and is fully connected to C5.It has 10164trainable parameters.As in classical NN’s,units in layers up to F6compute a dot product between their input vector and their weight vector,to which a bias is added.This weighted sum,denotedforunit (6)wheredeterminesits slope at the origin.Thefunctionis chosen to be1.7159.The rationale for this choice of a squashing function is given in Appendix A.Finally,the output layer is composed of Euclidean RBF units,one for each class,with 84inputs each.The outputs of each RBFunit(7)In other words,each output RBF unit computes the Eu-clidean distance between its input vector and its parameter vector.The further away the input is from the parameter vector,the larger the RBF output.The output of a particular RBF can be interpreted as a penalty term measuring the fit between the input pattern and a model of the class associated with the RBF.In probabilistic terms,the RBF output can be interpreted as the unnormalized negative log-likelihood of a Gaussian distribution in the space of configurations of layer F6.Given an input pattern,the loss function should be designed so as to get the configuration of F6as close as possible to the parameter vector of the RBF that corresponds to the pattern’s desired class.The parameter vectors of these units were chosen by hand and kept fixed (at least initially).The components of thoseparameters vectors were set to1.While they could have been chosen at random with equal probabilities for1,or even chosen to form an error correctingcode as suggested by [47],they were instead designed to represent a stylized image of the corresponding character class drawn on a7。
基于BP神经网络的手写数字识别
BP训练过程
第一阶段，向前传播阶段： a）从样本集中取一个样本(X,Yp)，X是输入向量，Yp是理想输出向量，将X输入网络； b）计算相应的实际输出Op。
在此阶段，信息从输入层经过逐级的变换，传送到输出层。这个过程也是网络在完成 训练后正常运行时执行的过程。在此过程中，网络执行的是计算（实际上就是输入与每层的 权值矩阵相点乘，得到最后的输出结果）：
ZIPCODE RECOGNITION
选择手写数字识别作为研究对象是因为这是一个相对简单的机器视觉任务： 1.将黑白像素点作为输入； 2.数字能够很好地与背景分离开； 3.输出只有10个类别；
用最简单的神经网络进行识别28*28像素的图片
存在的问题： 1. 一般要得到较好的训练效果，隐层数目不能太少，当图 片大的时候，需要的权值会非常多！ 2. 对平移、尺度变化敏感（比如数字偏左上角，右下角时 即识别失败） 3. 图片在相邻区域是相关的，而这种网络只是一股脑把所 有像素扔进去，没有考虑图片相关性。
LeNet-5手写识别系统
6个5X5模板
LeNet-5共有7层，不包含输入，每层都包含可训练参数（连接权重）。输入图像为32*32大小。 这要比Mnist数据库（一个公认的手写数据库）中最大的字母还大。这样做的原因是希望潜在 的明显特征如笔画断电或角点能够出现在最高层特征监测子感受野的中心
每个层有多个Feature Map，每个Feature Map通过一种卷积滤波器提取输入的一种特征 (每种特征都不一样)，然后每个Feature Map有多个神经元。
PREPROCESSING
在字符识别的过程中，识别算法不需要关心图像的彩色信息。因此，需要将彩色图像转化为灰度图像。 经过灰度化处理后的图像中还包含有背景信息。因此，我们还得进一步处理，将背景噪声屏蔽掉，突显出 字符轮廓信息。二值化处理就能够将其中的字符显现出来，并将背景去除掉。
神经网络 论文
神经网络论文以下是一些关于神经网络的重要论文：1. "A Computational Approach to Edge Detection"，作者：John Canny，论文发表于1986年，提出了一种基于神经网络的边缘检测算法，被广泛应用于计算机视觉领域。
2. "Backpropagation Applied to Handwritten Zip Code Recognition"，作者：Yann LeCun et al.，论文发表于1990年，引入了反向传播算法在手写数字识别中的应用，为图像识别领域开创了先河。
3. "Gradient-Based Learning Applied to Document Recognition"，作者：Yann LeCun et al.，论文发表于1998年，介绍了LeNet-5，一个用于手写数字和字符识别的深度卷积神经网络。
4. "ImageNet Classification with Deep Convolutional Neural Networks"，作者：Alex Krizhevsky et al.，论文发表于2012年，提出了深度卷积神经网络模型（AlexNet），在ImageNet图像识别竞赛中取得了重大突破。
5. "Deep Residual Learning for Image Recognition"，作者：Kaiming He et al.，论文发表于2015年，提出了深度残差网络（ResNet），通过引入残差连接解决了深度神经网络训练中的梯度消失和梯度爆炸问题。
6. "Generative Adversarial Networks"，作者：Ian Goodfellow etal.，论文发表于2014年，引入了生成对抗网络（GAN），这是一种通过博弈论思想训练生成模型和判别模型的框架，广泛应用于图像生成和增强现实等领域。
(中文)零基础深度学习deep learning
目录[1] Deep learning简介[2] Deep Learning训练过程[3] CNN卷积神经网络推导和实现[4] CNN的反向求导及练习[5] CNN卷积神经网络（一）深度解析CNN[6] CNN卷积神经网络（二）文字识别系统LeNet-5[7] CNN卷积神经网络（三）CNN常见问题总结[1] Deep learning简介一、什么是Deep Learning？实际生活中，人们为了解决一个问题，如对象的分类（对象可是是文档、图像等），首先必须做的事情是如何来表达一个对象，即必须抽取一些特征来表示一个对象，如文本的处理中，常常用词集合来表示一个文档，或把文档表示在向量空间中（称为VSM 模型），然后才能提出不同的分类算法来进行分类；又如在图像处理中，我们可以用像素集合来表示一个图像，后来人们提出了新的特征表示，如SIFT，这种特征在很多图像处理的应用中表现非常良好，特征选取得好坏对最终结果的影响非常巨大。
因此，选取什么特征对于解决一个实际问题非常的重要。
然而，手工地选取特征是一件非常费力、启发式的方法，能不能选取好很大程度上靠经验和运气；既然手工选取特征不太好，那么能不能自动地学习一些特征呢？答案是能！Deep Learning就是用来干这个事情的，看它的一个别名Unsupervised Feature Learning，就可以顾名思义了，Unsupervised的意思就是不要人参与特征的选取过程。
因此，自动地学习特征的方法，统称为Deep Learning。
二、Deep Learning的基本思想假设我们有一个系统S，它有n层（S1,…Sn），它的输入是I，输出是O，形象地表示为：I =>S1=>S2=>…..=>Sn => O，如果输出O等于输入I，即输入I经过这个系统变化之后没有任何的信息损失（呵呵，大牛说，这是不可能的。
信息论中有个“信息逐层丢失”的说法（信息处理不等式），设处理a信息得到b，再对b处理得到c，那么可以证明：a和c的互信息不会超过a和b的互信息。
浙江省台州市英语高三上学期期中试卷及解答参考(2024年)
2024年浙江省台州市英语高三上学期期中自测试卷及解答参考一、听力第一节（本大题有5小题，每小题1.5分，共7.5分）1、What does the man mean when he says “I’m afraid it’s impossible to finish the project on time”?A. He is worried about the project’s deadline.B. He is confident that the project will be completed on time.C. He is unable to complete the project himself.D. He is not interested in working on the project.Answer: AExplanation: The phrase “I’m afraid” indicates concern or doubt, and “it’s impossib le to finish the project on time” shows that the man believes there is a risk of not meeting the deadline. Therefore, the correct answer is A.2、Why does the woman say she needs to go to the library?A. She wants to borrow some books.B. She needs to find a quiet place to study.C. She needs to research information for a school project.D. She needs to return some books she borrowed.Answer: CExplanation: The woman mentions “a school project” as the reason for needing to go to the library, which implies that she requires information for her project. Therefore, the correct answer is C.3、Conversation:Man:I can’t believe it’s already October. Time flies!Woman:Yeah, and before we know it, it’s going to be Christmas. I need to start planning for the family gathering.Question: What does the woman imply she needs to do soon?A)Go on a vacation.B)Start preparing for Christmas.C)Visit her family.D)Buy a new calendar.Answer: B) Start preparing for Christmas.Explanation: The woman mentions that “before we know it, it’s going to be Christmas” and then states that she “needs to start planning for the family gathering,” which implies that she is thinking about preparations for the upcoming Christmas holiday.4、Conversation:Woman:Did you see that our team won the championship? It was such an exciting match!Man:I wish I could have been there. I heard it went into overtime. Who scored the winning goal?Woman:It was Sarah, with just seconds left on the clock. Everyone went wild!Question: According to the woman, what happened at the end of the game?A)The man arrived late to the game.B)The game ended in a tie.C)Sarah scored the winning goal.D)The team lost the championship.Answer: C) Sarah scored the winning goal.Explanation: The woman explains that Sarah scored the winning goal with only seconds remaining, leading to the team’s victory and excitement among the spectators. This directly points to option C as the correct response to the question.5.How many books did the librarian recommend to the student last week?A. 3B. 4C. 5D. 6Answer: AExplanation: The librarian mentioned in the conversation that she recommended three books to the student last week. Therefore, the correct answer is A. 3.二、听力第二节（本大题有15小题，每小题1.5分，共22.5分）1、Listen to the following conversation between two students discussing their plans forthe weekend.ConversationStudent A: Hey, I was thinking about going to the beach this weekend. The weather forecast says it’s going to be sunny.Student B: That sounds great, but didn’t you say you had a lot of homework to finish?Student A: Yeah, but it can wait until Monday. I think we should take advantage of the nice weather while we can.Questions1、What does Student A suggest doing on the weekend?Answer: Going to the beach.Explanation: From the conversation, it’s clear that Student A suggests takinga trip to the beach over the weekend because the weather is expected to be nice.2、Why does Student B seem hesitant about the plan?Answer: Stud ent B is concerned about Student A’s unfinished homework. Explanation: Student B reminds Student A about having a lot of homework, indicating concern that going to the beach might delay completing schoolwork.End of Listening Section Part II Questions 1-2Please wait for further instructions before continuing with the next set of questions.3、You will hear a short conversation between two students about their weekend plans. Listen carefully and answer the question.Question: What does the girl suggest doing for their weekend activity?A. Going to the movies.B. Visiting a museum.C. Going on a hike.D. Staying at home.Answer: CExplanation: The girl suggests going on a hike because she mentions, “I thought we could go hiking this weekend. It’s so beautiful outside.”4、You will hear a news report about a recent event in the city. Listen carefully and answer the question.Question: What is the main purpose of the news report?A. To inform about a new traffic rule.B. To discuss the city’s upcoming events.C. To report a major traffic accident.D. To announce the winner of a local competition.Answer: CExplanation: The news report focuses on a major traffic accident that occurred recently in the city, detailing the aftermath and any ongoing investigations.5、What does the man suggest they do first?A. Go to the movies.B. Have dinner at a restaurant.C. Visit the new art exhibition in town.Answer: CExplanation: In the conversation, the man suggests that they should visit the new art exhibition in town before it gets too crowded, implying that this would be their first activity.Conversation 2Script not provided, as this is an audio-based question.6、Why is the woman going to New York?A. To attend a business conference.B. To visit her family.C. To go on vacation.Answer: AExplanation: The woman mentions that she has a business conference in New York, which is the main reason for her trip. She plans to stay a few extra days after the conference to explore the city, but the primary purpose of her visit is work-related.7.W: I can’t believe it’s already November. Time flies!M: Yeah, it really does. I can’t remember what we were talking about last month.Q: What does the man imply about time?A) It is going by too quickly.B) He can’t remember what happened last month.C) It seems like it’s only been a few weeks.D) November is his favorite month.Answer: A) It is going by too quickly.Explanation: The man’s comment “Time flies!” indicates that he feel s time is passing very quickly.8.M: Did you finish reading that novel I lent you last week?W: Not yet, but I’m almost halfway through. It’s really gripping and I can’t put it down.Q: What does the woman mean when she says “I can’t put it down”?A) She is losing her place in the book.B) She is too tired to continue reading.C) She is finding the book difficult to understand.D) She is so interested in the book that she doesn’t want to stop reading.Answer: D) She is so interested in the book that she doesn’t want to stop reading.Explanation: The phrase “I can’t put it down” is a common expression that means someone is so engrossed in something, like a book, that they can’t stop doing it.9、According to the passage, what is the main initiative the community has undertaken to promote environmental awareness?A) Organizing weekly recycling drivesB) Starting a tree planting campaignC) Hosting monthly educational seminarsD) Implementing a ban on single-use plasticsAnswer: DExplanation: The passage clearly states that the most significant step taken by the community was to implement a ban on all single-use plastics, which has had a positive impact on reducing waste.10、What additional measure does the speaker suggest could further improve the commun ity’s environmental efforts?A) Introducing fines for litteringB) Encouraging carpooling to reduce emissionsC) Developing a community gardenD) Increasing public transportation optionsAnswer: AExplanation: While the speaker mentions several good practices already in place, they specifically suggest that introducing fines for littering could be an effective way to deter people from improperly disposing of waste and could complement existing initiatives.11.You will hear a conversation between two students discussing their study plans. Listen to the conversation and answer the following question.Question: How does the student feel about the upcoming midterm exam?A. ExcitedB. NervousC. RelaxedD. IndifferentAnswer: B. NervousExplanation: In the conversation, one student mentions, “I’m really worried about the midterm exam. There’s so much material to cover.” This indicates that the student feels nervous about the upcoming exam.12.You will hear a short lecture about the importance of teamwork in group projects. Listen to the lecture and answer the following question.Question: What is the main idea of the lecture?A. Individual work is more important than teamwork.B. Teamwork is essential for successful group projects.C. Group projects are unnecessary in high school.D. Team projects are only beneficial for students in certain subjects.Answer: B. Teamwork is essential for successful group projects.Explanation: The lecture emphasizes the importance of teamwork in group projec ts. The speaker states, “Teamwork is crucial for successful group projects because it allows students to share ideas and learn from each other.” This supports option B as the main idea of the lecture.13、What is the main topic of the passage?A) The importance of reducing pollution in rural areas.B) Urban initiatives aimed at protecting the environment.C) Strategies for conserving energy in industrial zones.D) The role of government policies in forest preservation.Answer: B) Urban initiatives aimed at protecting the environment.Explanation: The passage discusses several projects in cities that aim to reduce waste, promote recycling, and create green spaces, indicating that the focus is on environmental conservation within urban settings.14、According to the speaker, what is one of the benefits of planting more trees in city centers?A) It increases property values significantly.B) It reduces noise levels from traffic.C) It provides habitats for wildlife.D) It leads to improved air quality.Answer: D) It leads to improved air quality.Explanation: During the passage, the speaker mentions that increasing the number of trees in city centers helps to combat air pollution by absorbing carbon dioxide and releasing oxygen, thus contributing to better air quality.15.Listen to the following conversation and choose the best answer to the question.A. The man is going to visit his friend in the next few days.B. The woman is planning a trip to the countryside.C. They are discussing the man’s upcoming birthday party.D. The woman is trying to persuade the man to go to the gym.Answer: AExplanation: In the conversation, the woman asks the man if he is planningto visit his friend in the next few days. The man confirms that he is, so the correct answer is option A.三、阅读第一节（第1题7.5分，其余每题10分，总37.5分）First QuestionPassage:The Importance of Reading in the Digital AgeIn an era dominated by screens, from smartphones to tablets and computers, the act of reading has undergone a significant transformation. While some argue that traditional books are becoming obsolete, others believe that the value of physical books remains unparalleled. The debate over digital versus print media has sparked numerous discussions among educators, students, and book lovers alike.One advantage of digital reading is its accessibility. With e-books, readers can carry thousands of titles on a single device, making literature more accessible than ever before. Moreover, e-books often come with features like adjustable font sizes, search functions, and instant access to definitions or translations, which can enhance the reading experience for many users.However, proponents of traditional reading argue that there is something special about holding a book, turning its pages, and feeling its weight. They contend that the tactile experience cannot be replicated by electronic devices. Additionally, concerns have been raised about the potential negative effectsof prolonged screen time on eye health and sleep patterns.Ultimately, whether one chooses to read digitally or through traditional means comes down to personal preference. Both methods offer unique benefits and contribute to the richness of our literary experiences. As technology continues to evolve, it is likely that we will see further integration between digital and print media, creating new opportunities for readers to engage with written content.1、According to the passage, what is an advantage of e-books over traditional books?Answer: The accessibility of e-books; readers can carry thousands of titles on a single device.2、What concern is mentioned regarding digital reading?Answer: Potential negative effects of prolonged screen time on eye health and sleep patterns.3、What does the passage suggest about the future of reading?Answer: Further integration between digital and print media, creating new opportunities for engagement with written content.4、Which of the following is NOT mentioned as a feature of e-books?Answer: Automatic bookmarking (This is not mentioned in the passage as a feature of e-books).第二题Passage:In recent years, the rise of social media has significantly changed the way people communicate and share information. Platforms like Facebook, Twitter, and Instagram have become integral parts of daily life for millions around the world. While these platforms offer numerous benefits, such as easy access to news, entertainment, and social connections, they also come with a range of negative consequences.One major concern is the impact on mental health. Studies have shown that excessive use of social media can lead to feelings of loneliness, anxiety, and depression. The constant exposure to curated images and lifestyles can create unrealistic expectations and a sense of inadequacy. Additionally, the immediate feedback loop provided by social media can exacerbate negative emotions and behaviors.Another issue is the spread of misinformation. With the ease of sharing content, false news and rumors can spread rapidly, often without beingfact-checked. This not only undermines the credibility of legitimate news sources but also has serious implications for public discourse and political stability.Despite these challenges, social media also provides valuable opportunities for education and empowerment. Online platforms have become powerful tools for activism, allowing individuals to raise awareness about social issues and mobilize for change. They also offer a space for learning and self-improvement, with countless educational resources and communities available at the click ofa button.Questions:1、What is one of the negative consequences mentioned in the passage regarding social media usage?A) Improved communication skillsB) Increased social connectionsC) Feelings of loneliness, anxiety, and depressionD) Enhanced self-confidence2、What is the main concern regarding the spread of misinformation on social media?A) Increased access to educational resourcesB) The rapid spread of false news and rumorsC) Improved political stabilityD) Enhanced credibility of news sources3、According to the passage, how can social media be a positive influence?A) By promoting false news and rumorsB) By encouraging feelings of inadequacyC) By providing a space for activism and learningD) By creating unrealistic expectations4、What is the author’s overall tone towards social media in the passage?A) PessimisticB) OptimisticC) NeutralD) CriticalAnswers:1、C) Feelings of loneliness, anxiety, and depression2、B) The rapid spread of false news and rumors3、C) By providing a space for activism and learning4、D) Critical第三题Read the following passage and answer the questions that follow.In the small coastal town of Seaview, the annual Seaview International Film Festival has become a major event that brings filmmakers, actors, and film enthusiasts from around the world. The festival, which takes place every October, celebrates the art of cinema and showcases a diverse range of films, from documentaries to feature films. The festival also includes workshops, panel discussions, and film screenings that cater to both professionals and amateurs.The festival’s origins date back to the early 2000s when a group of local film enthusiasts decided to organize a small film screening event in the local community center. Over the years, the event has grown exponentially, attracting international attention and support. The festival’s success can be attributed to its commitment to promoting independent cinema and providing a platform for emerging filmmakers to showcase their work.One of the highlights of the festival is the “Emerging Filmmakers Showcase,”which features short films from up-and-coming directors. This section of the festival has gained a reputation for discovering new talents and has even led to some filmmakers securing distribution deals for their films.This year, the festival is celebrating its 15th anniversary, and the organizers have planned a special edition that includes a retrospective of some of the most memorable films from the past decade. The festival will also honor a lifetime achievement award to a renowned filmmaker who has made significant contributions to the film industry.1、What is the main purpose of the Seaview International Film Festival?A. To promote local tourismB. To celebrate the art of cinema and support emerging filmmakersC. To showcase the best-selling films of the yearD. To provide workshops for film professionals2、How did the Seaview International Film Festival begin?A. It was established by a renowned filmmaker.B. It started as a small film screening event organized by local film enthusiasts.C. It was created to honor a lifetime achievement award recipient.D. It was founded to promote independent cinema from around the world.3、What is a notable feature of the “Emerging Filmmakers Showcase”?A. It showcases only feature films.B. It is known for its high entry fees.C. It has a strict selection process.D. It has led to some filmmakers securing distribution deals for their films.4、What special event is planned for the festival’s 15th anniversary?A. A retrospective of the most memorable films from the past decade.B. A panel discussion on the future of independent cinema.C. An exclusive screening of the latest film from a renowned filmmaker.D. A competition for the best short film from emerging filmmakers.Answers:1、B2、B3、D4、A第四题Reading Passage:In the small coastal town of Seaview, the annual Seaview Festival is a much-anticipated event that brings together local residents and tourists from far and wide. The festival celebrates the town’s rich maritime history and its vibrant culture. This year, the festival is scheduled to take place over a three-day period, from November 10th to November 12th.The festival kicks off with a parade through the town’s main street, where participants wear traditional costumes and display local crafts. The parade is followed by a series of workshops and demonstrations, including ship-building,knot-tying, and marine biology. The highlight of the festival, however, is the Seafood Festival, where visitors can taste a variety of fresh seafood dishes prepared by local chefs.1、What is the main purpose of the Seaview Festival?A. To promote tourism in the town.B. To celebrate the town’s maritime history.C. To showcase the town’s cultural diversity.D. To raise funds for local charities.2、On which day does the Seaview Festival begin?A. November 9thB. November 10thC. November 11thD. November 12th3、Which of the following activities is NOT mentioned as part of the festival’s workshops and demonstrations?A. Ship-buildingB. Cooking classesC. Marine biologyD. Pottery making4、What is the main attraction of the Seafood Festival?A. A fishing competitionB. A cooking competitionC. A seafood festival with various dishesD. A seafood market where visitors can buy their own seafoodAnswers:1、B2、B3、D4、C四、阅读第二节（12.5分）Title: The Benefits of Physical Exercise for StudentsPhysical exercise is not only beneficial for adults but also crucial for students’ physical and mental development. In today’s fast-paced world, students often spend a significant amount of time sitting in classrooms, studying, and using electronic devices. This sedentary lifestyle can lead to various health issues, such as obesity, poor posture, and stress. Therefore, incorporating physical exercise into students’ daily routines is essential.Regular physical exercise helps students maintain a healthy weight. According to the Centers for Disease Control and Prevention (CDC), approximately 17% of children and adolescents aged 2 to 19 years are obese. Engaging in physical activities can help burn excess calories, reducing the risk of obesity.Moreover, physical exerci se improves students’ posture. Sitting for extended periods can lead to poor posture, which may result in back pain andother health issues. Activities like yoga and Pilates can help strengthen the muscles supporting the spine, thus improving posture.Phys ical exercise also has a positive impact on students’ mental health. Exercise releases endorphins, the body’s natural painkillers, which can help reduce stress and improve mood. Students who participate in regular physical activities are more likely to experience better academic performance and social skills.One study conducted by the University of California, Irvine, found that students who engaged in physical exercise for at least 20 minutes a day had higher grades and better attendance compared to those who did not exercise regularly.In addition to these benefits, physical exercise helps students develop discipline and teamwork skills. Participating in sports or group fitness classes can teach students the value of commitment, perseverance, and collaboration.However, it is essential to note that the type of physical activity should be appropriate for the age and fitness level of the students. For example, young children may enjoy playing outdoor games, while older students may prefer more intense activities like running or cycling.In conclusion, physical exercise plays a vital role in students’ overall well-being. It is crucial for schools and parents to encourage students to incorporate physical activity into their daily routines to reap the numerous benefits it offers.Questions:1.What is the main purpose of the article?A. To explain the causes of obesity in studentsB. To discuss the benefits of physical exercise for studentsC. To analyze the impact of sedentary lifestyles on studentsD. To propose solutions to improve students’ physical health2.According to the article, what percentage of children and adolescents aged2 to 19 years are obese?A. 10%B. 15%C. 17%D. 20%3.How does physical exercise help students improve their posture?A. It strengthens the muscles supporting the spineB. It reduces the risk of back painC. It improves the students’ overall fitnessD. It increases the students’ flexibility4.Which of the following is NOT a benefit of physical exercise mentioned in the article?A. Improved academic performanceB. Enhanced social skillsC. Reduced stressD. Increased obesity5.What is the author’s opinion on the appropriate type of physical activity for students?A. It should be determined by the students’ interestsB. It should be tailored to the students’ age and fitness levelC. It should be limited to outdoor activitiesD. It should be performed in group settings onlyAnswers:1.B2.C3.A4.D5.B五、语言运用第一节 _ 完形填空（15分）Title: English High School Grade 3 Semester 1 Midterm ExamV. Language ApplicationSection 1: Cloze TestRead the following passage and choose the best word for each blank from the options given below the passage.After spending the summer volunteering in a local community center,26-year-old Sarah had 27 a deep appreciation for the value of community service. She realized that it wasn’t just about helping others; it was also about 28personal growth and learning.Sarah’s experience began when she 29 at the community center to help with a summer reading program for children. The first day, she was 30 to a small, dimly lit room filled with eager young faces. Sarah introduced herself and the program, and she was 31 to see how excited the children were. Over the next few weeks, Sarah 32 stories, played games, and organized activities to keep the children engaged. She also had the opportunity to 33 with the parents, who were 34 to share their concerns and ideas.One day, Sarah 35 a young boy named Alex who seemed particularly quiet and withdrawn. She noticed that he didn’t 36 with the other children very much. Sarah decided to sit with Alex and 37 a book that he might be interested in. As they read together, she 38 that he had a passion for science and 39. She encouraged him to join a science club at the center, and he 40.Sarah’s summer volunteering experience was 41. She learned about the challenges faced by children from different backgrounds and the importance of patience and understanding. Most importantly, she realized that her own life had been 42 by the act of giving back to her community.Sarah’s story inspired many others to consider volunteering. She 43 that everyone has the potential to make a positive impact, no matter how small the effort.Choose the best word for each blank from the options below:26.A) developed B) gained C) achieved D) received27.A) gained B) achieved C) developed D) increased28.A) fostering B) encouraging C) promoting D) enhancing29.A) applied B) registered C) volunteered D) enrolled30.A) greeted B) welcomed C) introduced D) encountered31.A) surprised B) pleased C) disappointed D) scared32.A) read B) shared C) wrote D) presented33.A) communicate B) interact C) consult D) discuss34.A) excited B) interested C) concerned D) confident35.A) encountered B) noticed C) approached D) invited36.A) communicate B) interact C) socialize D) engage37.A) recommend B) suggest C) offer D) provide38.A) realized B) discovered C) acknowledged D) admitted39.A) skills B) interests C) knowledge D) experiences40.A) declined B) agreed C) hesitated D) refused41.A) inspiring B) successful C) frustrating D) exhausting42.A) affected B) influenced C) motivated D) inspired43.A) believed B) convinced C) realized D) understoodAnswer: 26. B) gained 27. A) gained 28. D) enhancing 29. C) volunteered 30.D) encountered 31. B) pleased 32. A) read 33. B) interact 34. C) concerned 35.B) noticed 36. C) socialize 37. A) recommend 38. B) discovered 39. B) interests40. B) agreed 41. A) inspiring 42. B) influenced 43. C) realized六、语言运用第二节 _ 语法填空（15分）Grammar CompletionRead the following passage and complete the sentences by choosing the most appropriate words or phrases from the options provided.Passage:The ancient city of Petra, located in Jordan, is a UNESCO World Heritage site and is renowned for its stunning rock-cut architecture. It was founded around the 6th century BC and has been inhabited by various civilizations over the centuries. Petra’s most famou s structure, Al Khazneh (the Treasury), is carved into the cliffs and is a masterpiece of ancient engineering. The city is also home to the Siq, a narrow passage that leads to the Treasury, and numerous other tombs and buildings. Visitors to Petra are often fascinated by the intricate details and the stories behind each structure. Despite its historical significance, the site faces challenges such as environmental degradation and tourism overdevelopment.Grammar Completion Questions:1.Petra’s stunning rock-cut architecture is a__________of ancient engineering.a) exampleb) resultc) processd) explanationAnswer:。
统计学习理论导论(清华大学张学工讲义)-1
• How to decide the structure of the MLP?
(How many hidden layers and nodes?)
– Ask God, or guess then pray
• How to choose the neuron function?
– Usually Sigmoid (S-shaped) function
– the effort to approach mathematic models for natural nervous systems
– the effort to implement man-made intelligence
• Three types of NN:
– Feedforward NN – Feedback NN – Competitive Learning (Self-organizing) NN
Xuegong Zhang
27
Tsinghua University
学习过程的应用分析与理论分析学派
• 关于感知器学习能力的若干结论： – 关于收敛性的结论 – 关于收敛以后的测试错误率（推广能力）的结论
[Novikoff, 1962] [Aizerman, Braverman, and Rozonoer, 1964]
• 学习过程的应用分析学派：
– 最小化训练错误数是不言而喻的归纳原则，学习的主要问题在于 寻找同时构造所有神经元的系数的方法，使所形成的分类面能达
到最小的训练错误率，（这样即可得到好的推广性）
• 学习过程的理论分析学派：
Xuegong Zhang
14
Tsinghua University
Handwritten Digit Recognition with a Back-propagation Network
The main point of this paper is to show that large back-propagation (BP) networks can be applied to real image-recognition problems without a large, complex preprocessing stage requiring detailed engineering. Unlike most previous work on the subject (Denker et al., 1989), the learning network is directly fed with images, rather than feature vectors, thus demonstrating the ability of BP networks to deal with large amounts of low level information. Previous work performed on simple digit images (Le Cun, 1989) showed that the architecture of the network strongly in uences the network's generalization ability. Good generalization can only be obtained by designing a network architecture that contains a certain amount of a priori knowledge about the problem. The basic design principle is to minimize the number of free parameters that must be determined by the learning algorithm, without overly reducing the computational power of the network. This principle increases the probability of correct generalization because
分享两篇SCI发表的经历(cover letter、response letter)
分享两篇SCI发表的经历三年前对于我来说SCI就是天书一样，在我踏进博士的门槛后我以为自己进入了地狱，也纠结也彷徨，整天刷虫友们对于博士、SCI的帖子，我选择了虫友们鼓励的那一部分来激励自己继续前行。
我告诉自己坚持就是胜利，当然那是积极的坚持。
在好几月之前就有这个想法，今天早上收到第二篇的接收通知后，我便想今天一定要在小木虫上谢谢那些给予我帮助的虫友们。
话不多说，我把自己这两篇投稿的经历与大家共享，希望能给大家带来一点点用处。
第一篇发表在FitoterapiaCover letterDear Editor Verotta:We would like to submit the manuscript entitled "××××××题目" by ××××××所有作者姓名which we wish to be considered for publication in Journal of Fitoterapia.All authors have read and approved this version of the article, and due care has been taken to ensure the integrity of the work. Neither the entire paper nor any part of its content has been published or has been accepted elsewhere. It is not being submitted to any other journal.We believe the paper may be of particular interest to the readers of your journal as it is the first time of ××××××研究的精华所在Thank you very much for your reconsidering our revised manuscript for potential publication in Fitoterapia. We are looking forward to hearing from you soon. Correspondence should be addressed to Jinhui Yu at the following address, phone and fax number, and email address.地址、学院、学校名称Phone: + 86××××××Fax number: + 86571××××××Best wishes for you and your family!Sincerely yours,×××所有作者Response to reviewersDear Editor:Thank you very much for your letter and the comments from the referees about our paper submitted to Journal of Fitoterapia (FITOTE-D-11-01071). The manuscript entitled "××××××" by ××××××所有作者have been revised according to the reviewers’ comments, and we wish it to be reconsidered for publication in Journal of Fitoterapia.A list of changes and responses to reviewers are as follows.List of ActionsLOA1: The key words were changed in page?.LOA2: The name and location of the local biochemistry company have been added in section 2.1 (page 3).LOA3: A paragraph has been added in section 3.1 (page 5) to further explain the determination of the cis and trans configuration of double bonds in polyprenols.LOA4: The language was improved by English language editing of Elsevier webshop.To Reviewer 1#,Thank you very much for pointing out the problems in our manuscript. We have revised it according to your recommendations. We would like to know if there are still somewhere need to be amended.(1) Keywords: general terms should be avoided; I would change some of the keywords (homologues, identification, quantification)The key words have been changed as follows: ××××××修改后关键词(2) In paragraph 2.1 the "local biochemistry company" should be identified byname and location.The name and location of the local biochemistry company have been added in section 2.1 (page 3). NaOH，Pyrogallol，anhydrous Na2SO4 were purchased from Hangzhou ChangqingHuagong CO., LTD.(3) How the cis and trans configuration of double bonds in ××××××were determined? Authors should say something about.The following paragraph has been added in section 3.1 (page 5) to further explain the determination of the cis and trans configuration of double bonds in ××××××.(4) Language should be checked for clarity and correctness.The language was improved by English language editing of Elsevier webshop.To Reviewer 2#,Thank you very much for your recommendation on our paper and we have improved by English language editing of Elsevier webshop.All in all, thank you very much for your reconsidering our revised manuscript for potential publication in Fitoterapia. I'm looking forward to hearing from you soon. Correspondence should be addressed to ****第一作者或通讯作者at the following address, phone and fax number, and email address.地址Phone: + 86571××××××Fax number: + ××××××Best wishes for you and your family!Sincerely yours,××××××所有作者总的来说，第一篇文章没有费很大劲。
Convolutional Networks for Images Speech and Time Series
LeCun & Bengio: Convolutional Networks for Images, Speech, and Time-Series
3
1 INTRODUCTION
The ability of multilayer back-propagation networks to learn complex, high-dimensional, nonlinear mappings from large collections of examples makes them obvious candidates for image recognition or speech recognition tasks (see PATTERN RECOGNITION AND NEURAL NETWORKS). In the traditional model of pattern recognition, a hand-designed feature extractor gathers relevant information from the input and eliminates irrelevant variabilities. A trainable classi er then categorizes the resulting feature vectors (or strings of symbols) into classes. In this scheme, standard, fully-connected multilayer networks can be used as classi ers. A potentially more interesting scheme is to eliminate the feature extractor, feeding the network with \raw" inputs (e.g. normalized images), and to rely on backpropagation to turn the rst few layers into an appropriate feature extractor. While this can be done with an ordinary fully connected feed-forward network with some success for tasks such as character recognition, there are problems. Firstly, typical images, or spectral representations of spoken words, are large, often with several hundred variables. A fully-connected rst layer with, say a few 100 hidden units, would already contain several 10,000 weights. Over tting problems may occur if training data is scarce. In addition, the memory requirement for that many weights may rule out certain hardware implementations. But, the main de ciency of unstructured nets for image or speech aplications is that they have no built-in invariance with respect to translations, or
IEEE参考文献格式
•Creating a reference list or bibliographyA numbered list of references must be provided at the end of thepaper. The list should be arranged in the order of citation in the text of the assignment or essay, not in alphabetical order. List only one reference per reference number. Footnotes or otherinformation that are not part of the referencing format should not be included in the reference list.The following examples demonstrate the format for a variety of types of references. Included are some examples of citing electronic documents. Such items come in many forms, so only some examples have been listed here.Print DocumentsBooksNote: Every (important) word in the title of a book or conference must be capitalised. Only the first word of a subtitle should be capitalised. Capitalise the "v" in Volume for a book title.Punctuation goes inside the quotation marks.Standard formatSingle author[1] W.-K. Chen, Linear Networks and Systems. Belmont, CA: Wadsworth,1993, pp. 123-135.[2] S. M. Hemmington, Soft Science. Saskatoon: University ofSaskatchewan Press, 1997.Edited work[3] D. Sarunyagate, Ed., Lasers. New York: McGraw-Hill, 1996.Later edition[4] K. Schwalbe, Information Technology Project Management, 3rd ed.Boston: Course Technology, 2004.[5] M. N. DeMers, Fundamentals of Geographic Information Systems,3rd ed. New York : John Wiley, 2005.More than one author[6] T. Jordan and P. A. Taylor, Hacktivism and Cyberwars: Rebelswith a cause? London: Routledge, 2004.[7] U. J. Gelinas, Jr., S. G. Sutton, and J. Fedorowicz, Businessprocesses and information technology. Cincinnati:South-Western/Thomson Learning, 2004.Three or more authorsNote: The names of all authors should be given in the references unless the number of authors is greater than six. If there are more than six authors, you may use et al. after the name of the first author.[8] R. Hayes, G. Pisano, D. Upton, and S. Wheelwright, Operations,Strategy, and Technology: Pursuing the competitive edge.Hoboken, NJ : Wiley, 2005.Series[9] M. Bell, et al., Universities Online: A survey of onlineeducation and services in Australia, Occasional Paper Series 02-A. Canberra: Department of Education, Science andTraining, 2002.Corporate author (ie: a company or organisation)[10] World Bank, Information and Communication Technologies: AWorld Bank group strategy. Washington, DC : World Bank, 2002.Conference (complete conference proceedings)[11] T. J. van Weert and R. K. Munro, Eds., Informatics and theDigital Society: Social, ethical and cognitive issues: IFIP TC3/WG3.1&3.2 Open Conference on Social, Ethical andCognitive Issues of Informatics and ICT, July 22-26, 2002, Dortmund, Germany. Boston: Kluwer Academic, 2003.Government publication[12] Australia. Attorney-Generals Department. Digital AgendaReview, 4 Vols. Canberra: Attorney- General's Department,2003.Manual[13] Bell Telephone Laboratories Technical Staff, TransmissionSystem for Communications, Bell Telephone Laboratories,1995.Catalogue[14] Catalog No. MWM-1, Microwave Components, M. W. Microwave Corp.,Brooklyn, NY.Application notes[15] Hewlett-Packard, Appl. Note 935, pp. 25-29.Note:Titles of unpublished works are not italicised or capitalised. Capitalise only the first word of a paper or thesis.Technical report[16] K. E. Elliott and C.M. Greene, "A local adaptive protocol,"Argonne National Laboratory, Argonne, France, Tech. Rep.916-1010-BB, 1997.Patent / Standard[17] K. Kimura and A. Lipeles, "Fuzzy controller component, " U.S. Patent 14,860,040, December 14, 1996.Papers presented at conferences (unpublished)[18] H. A. Nimr, "Defuzzification of the outputs of fuzzycontrollers," presented at 5th International Conference onFuzzy Systems, Cairo, Egypt, 1996.Thesis or dissertation[19] H. Zhang, "Delay-insensitive networks," M.S. thesis,University of Waterloo, Waterloo, ON, Canada, 1997.[20] M. W. Dixon, "Application of neural networks to solve therouting problem in communication networks," Ph.D.dissertation, Murdoch University, Murdoch, WA, Australia, 1999.Parts of a BookNote: These examples are for chapters or parts of edited works in which the chapters or parts have individual title and author/s, but are included in collections or textbooks edited by others. If the editors of a work are also the authors of all of the included chapters then it should be cited as a whole book using the examples given above (Books).Capitalise only the first word of a paper or book chapter.Single chapter from an edited work[1] A. Rezi and M. Allam, "Techniques in array processing by meansof transformations, " in Control and Dynamic Systems, Vol.69, Multidemsional Systems, C. T. Leondes, Ed. San Diego: Academic Press, 1995, pp. 133-180.[2] G. O. Young, "Synthetic structure of industrial plastics," inPlastics, 2nd ed., vol. 3, J. Peters, Ed. New York:McGraw-Hill, 1964, pp. 15-64.Conference or seminar paper (one paper from a published conference proceedings)[3] N. Osifchin and G. Vau, "Power considerations for themodernization of telecommunications in Central and Eastern European and former Soviet Union (CEE/FSU) countries," in Second International Telecommunications Energy SpecialConference, 1997, pp. 9-16.[4] S. Al Kuran, "The prospects for GaAs MESFET technology in dc-acvoltage conversion," in Proceedings of the Fourth AnnualPortable Design Conference, 1997, pp. 137-142.Article in an encyclopaedia, signed[5] O. B. R. Strimpel, "Computer graphics," in McGraw-HillEncyclopedia of Science and Technology, 8th ed., Vol. 4. New York: McGraw-Hill, 1997, pp. 279-283.Study Guides and Unit ReadersNote: You should not cite from Unit Readers, Study Guides, or lecture notes, but where possible you should go to the original source of the information. If you do need to cite articles from the Unit Reader, treat the Reader articles as if they were book or journal articles. In the reference list or bibliography use the bibliographical details as quoted in the Reader and refer to the page numbers from the Reader, not the original page numbers (unless you have independently consulted the original).[6] L. Vertelney, M. Arent, and H. Lieberman, "Two disciplines insearch of an interface: Reflections on a design problem," in The Art of Human-Computer Interface Design, B. Laurel, Ed.Reading, MA: Addison-Wesley, 1990. Reprinted inHuman-Computer Interaction (ICT 235) Readings and Lecture Notes, Vol. 1. Murdoch: Murdoch University, 2005, pp. 32-37. Journal ArticlesNote: Capitalise only the first word of an article title, except for proper nouns or acronyms. Every (important) word in the title of a journal must be capitalised. Do not capitalise the "v" in volume for a journal article.You must either spell out the entire name of each journal that you reference or use accepted abbreviations. You must consistently do one or the other. Staff at the Reference Desk can suggest sources of accepted journal abbreviations.You may spell out words such as volume or December, but you must either spell out all such occurrences or abbreviate all. You do not need to abbreviate March, April, May, June or July.To indicate a page range use pp. 111-222. If you refer to only one page, use only p. 111.Standard formatJournal articles[1] E. P. Wigner, "Theory of traveling wave optical laser," Phys.Rev., vol. 134, pp. A635-A646, Dec. 1965.[2] J. U. Duncombe, "Infrared navigation - Part I: An assessmentof feasability," IEEE Trans. Electron. Devices, vol. ED-11, pp. 34-39, Jan. 1959.[3] G. Liu, K. Y. Lee, and H. F. Jordan, "TDM and TWDM de Bruijnnetworks and shufflenets for optical communications," IEEE Trans. Comp., vol. 46, pp. 695-701, June 1997.OR[4] J. R. Beveridge and E. M. Riseman, "How easy is matching 2D linemodels using local search?" IEEE Transactions on PatternAnalysis and Machine Intelligence, vol. 19, pp. 564-579, June 1997.[5] I. S. Qamber, "Flow graph development method," MicroelectronicsReliability, vol. 33, no. 9, pp. 1387-1395, Dec. 1993.[6] E. H. Miller, "A note on reflector arrays," IEEE Transactionson Antennas and Propagation, to be published.Electronic documentsNote:When you cite an electronic source try to describe it in the same way you would describe a similar printed publication. If possible, give sufficient information for your readers to retrieve the source themselves.If only the first page number is given, a plus sign indicates following pages, eg. 26+. If page numbers are not given, use paragraph or other section numbers if you need to be specific. An electronic source may not always contain clear author or publisher details.The access information will usually be just the URL of the source. As well as a publication/revision date (if there is one), the date of access is included since an electronic source may change between the time you cite it and the time it is accessed by a reader.E-BooksStandard format[1] L. Bass, P. Clements, and R. Kazman. Software Architecture inPractice, 2nd ed. Reading, MA: Addison Wesley, 2003. [E-book] Available: Safari e-book.[2] T. Eckes, The Developmental Social Psychology of Gender. MahwahNJ: Lawrence Erlbaum, 2000. [E-book] Available: netLibrary e-book.Article in online encyclopaedia[3] D. Ince, "Acoustic coupler," in A Dictionary of the Internet.Oxford: Oxford University Press, 2001. [Online]. Available: Oxford Reference Online, .[Accessed: May 24, 2005].[4] W. D. Nance, "Management information system," in The BlackwellEncyclopedic Dictionary of Management Information Systems,G.B. Davis, Ed. Malden MA: Blackwell, 1999, pp. 138-144.[E-book]. Available: NetLibrary e-book.E-JournalsStandard formatJournal article abstract accessed from online database[1] M. T. Kimour and D. Meslati, "Deriving objects from use casesin real-time embedded systems," Information and SoftwareTechnology, vol. 47, no. 8, p. 533, June 2005. [Abstract].Available: ProQuest, /proquest/.[Accessed May 12, 2005].Note: Abstract citations are only included in a reference list if the abstract is substantial or if the full-text of the article could not be accessed.Journal article from online full-text databaseNote: When including the internet address of articles retrieved from searches in full-text databases, please use the Recommended URLs for Full-text Databases, which are the URLs for the main entrance to the service and are easier to reproduce.[2] H. K. Edwards and V. Sridhar, "Analysis of software requirementsengineering exercises in a global virtual team setup,"Journal of Global Information Management, vol. 13, no. 2, p.21+, April-June 2005. [Online]. Available: Academic OneFile, . [Accessed May 31, 2005].[3] A. Holub, "Is software engineering an oxymoron?" SoftwareDevelopment Times, p. 28+, March 2005. [Online]. Available: ProQuest, . [Accessed May 23, 2005].Journal article in a scholarly journal (published free of charge on the internet)[4] A. Altun, "Understanding hypertext in the context of readingon the web: Language learners' experience," Current Issues in Education, vol. 6, no. 12, July 2003. [Online]. Available: /volume6/number12/. [Accessed Dec. 2, 2004].Journal article in electronic journal subscription[5] P. H. C. Eilers and J. J. Goeman, "Enhancing scatterplots withsmoothed densities," Bioinformatics, vol. 20, no. 5, pp.623-628, March 2004. [Online]. Available:. [Accessed Sept. 18, 2004].Newspaper article from online database[6] J. Riley, "Call for new look at skilled migrants," TheAustralian, p. 35, May 31, 2005. Available: Factiva,. [Accessed May 31, 2005].Newspaper article from the Internet[7] C. Wilson-Clark, "Computers ranked as key literacy," The WestAustralian, para. 3, March 29, 2004. [Online]. Available:.au. [Accessed Sept. 18, 2004].Internet DocumentsStandard formatProfessional Internet site[1] European Telecommunications Standards Institute, 揇igitalVideo Broadcasting (DVB): Implementation guidelines for DVBterrestrial services; transmission aspects,?EuropeanTelecommunications Standards Institute, ETSI TR-101-190,1997. [Online]. Available: . [Accessed:Aug. 17, 1998].Personal Internet site[2] G. Sussman, "Home page - Dr. Gerald Sussman," July 2002.[Online]. Available:/faculty/Sussman/sussmanpage.htm[Accessed: Sept. 12, 2004].General Internet site[3] J. Geralds, "Sega Ends Production of Dreamcast," ,para. 2, Jan. 31, 2001. [Online]. Available:/news/1116995. [Accessed: Sept. 12,2004].Internet document, no author given[4] 揂憀ayman抯?explanation of Ultra Narrow Band technology,?Oct.3, 2003. [Online]. Available:/Layman.pdf. [Accessed: Dec. 3, 2003].Non-Book FormatsPodcasts[1] W. Brown and K. Brodie, Presenters, and P. George, Producer, 揊rom Lake Baikal to the Halfway Mark, Yekaterinburg? Peking to Paris: Episode 3, Jun. 4, 2007. [Podcast television programme]. Sydney: ABC Television. Available:.au/tv/pekingtoparis/podcast/pekingtoparis.xm l. [Accessed Feb. 4, 2008].[2] S. Gary, Presenter, 揃lack Hole Death Ray? StarStuff, Dec. 23, 2007. [Podcast radio programme]. Sydney: ABC News Radio. Available: .au/newsradio/podcast/STARSTUFF.xml. [Accessed Feb. 4, 2008].Other FormatsMicroform[3] W. D. Scott & Co, Information Technology in Australia:Capacities and opportunities: A report to the Department ofScience and Technology. [Microform]. W. D. Scott & CompanyPty. Ltd. in association with Arthur D. Little Inc. Canberra:Department of Science and Technology, 1984.Computer game[4] The Hobbit: The prelude to the Lord of the Rings. [CD-ROM].United Kingdom: Vivendi Universal Games, 2003.Software[5] Thomson ISI, EndNote 7. [CD-ROM]. Berkeley, Ca.: ISIResearchSoft, 2003.Video recording[6] C. Rogers, Writer and Director, Grrls in IT. [Videorecording].Bendigo, Vic. : Video Education Australasia, 1999.A reference list: what should it look like?The reference list should appear at the end of your paper. Begin the list on a new page. The title References should be either left justified or centered on the page. The entries should appear as one numerical sequence in the order that the material is cited in the text of your assignment.Note: The hanging indent for each reference makes the numerical sequence more obvious.[1] A. Rezi and M. Allam, "Techniques in array processing by meansof transformations, " in Control and Dynamic Systems, Vol.69, Multidemsional Systems, C. T. Leondes, Ed. San Diego: Academic Press, 1995, pp. 133-180.[2] G. O. Young, "Synthetic structure of industrial plastics," inPlastics, 2nd ed., vol. 3, J. Peters, Ed. New York:McGraw-Hill, 1964, pp. 15-64.[3] S. M. Hemmington, Soft Science. Saskatoon: University ofSaskatchewan Press, 1997.[4] N. Osifchin and G. Vau, "Power considerations for themodernization of telecommunications in Central and Eastern European and former Soviet Union (CEE/FSU) countries," in Second International Telecommunications Energy SpecialConference, 1997, pp. 9-16.[5] D. Sarunyagate, Ed., Lasers. New York: McGraw-Hill, 1996.[8] O. B. R. Strimpel, "Computer graphics," in McGraw-HillEncyclopedia of Science and Technology, 8th ed., Vol. 4. New York: McGraw-Hill, 1997, pp. 279-283.[9] K. Schwalbe, Information Technology Project Management, 3rd ed.Boston: Course Technology, 2004.[10] M. N. DeMers, Fundamentals of Geographic Information Systems,3rd ed. New York: John Wiley, 2005.[11] L. Vertelney, M. Arent, and H. Lieberman, "Two disciplines insearch of an interface: Reflections on a design problem," in The Art of Human-Computer Interface Design, B. Laurel, Ed.Reading, MA: Addison-Wesley, 1990. Reprinted inHuman-Computer Interaction (ICT 235) Readings and Lecture Notes, Vol. 1. Murdoch: Murdoch University, 2005, pp. 32-37.[12] E. P. Wigner, "Theory of traveling wave optical laser,"Physical Review, vol.134, pp. A635-A646, Dec. 1965.[13] J. U. Duncombe, "Infrared navigation - Part I: An assessmentof feasibility," IEEE Transactions on Electron Devices, vol.ED-11, pp. 34-39, Jan. 1959.[14] M. Bell, et al., Universities Online: A survey of onlineeducation and services in Australia, Occasional Paper Series 02-A. Canberra: Department of Education, Science andTraining, 2002.[15] T. J. van Weert and R. K. Munro, Eds., Informatics and theDigital Society: Social, ethical and cognitive issues: IFIP TC3/WG3.1&3.2 Open Conference on Social, Ethical andCognitive Issues of Informatics and ICT, July 22-26, 2002, Dortmund, Germany. Boston: Kluwer Academic, 2003.[16] I. S. Qamber, "Flow graph development method,"Microelectronics Reliability, vol. 33, no. 9, pp. 1387-1395, Dec. 1993.[17] Australia. Attorney-Generals Department. Digital AgendaReview, 4 Vols. Canberra: Attorney- General's Department, 2003.[18] C. Rogers, Writer and Director, Grrls in IT. [Videorecording].Bendigo, Vic.: Video Education Australasia, 1999.[19] L. Bass, P. Clements, and R. Kazman. Software Architecture inPractice, 2nd ed. Reading, MA: Addison Wesley, 2003. [E-book] Available: Safari e-book.[20] D. Ince, "Acoustic coupler," in A Dictionary of the Internet.Oxford: Oxford University Press, 2001. [Online]. Available: Oxford Reference Online, .[Accessed: May 24, 2005].[21] H. K. Edwards and V. Sridhar, "Analysis of softwarerequirements engineering exercises in a global virtual team setup," Journal of Global Information Management, vol. 13, no. 2, p. 21+, April-June 2005. [Online]. Available: AcademicOneFile, . [Accessed May 31,2005].[22] A. Holub, "Is software engineering an oxymoron?" SoftwareDevelopment Times, p. 28+, March 2005. [Online]. Available: ProQuest, . [Accessed May 23, 2005].[23] H. Zhang, "Delay-insensitive networks," M.S. thesis,University of Waterloo, Waterloo, ON, Canada, 1997.[24] P. H. C. Eilers and J. J. Goeman, "Enhancing scatterplots withsmoothed densities," Bioinformatics, vol. 20, no. 5, pp.623-628, March 2004. [Online]. Available:. [Accessed Sept. 18, 2004].[25] J. Riley, "Call for new look at skilled migrants," TheAustralian, p. 35, May 31, 2005. Available: Factiva,. [Accessed May 31, 2005].[26] European Telecommunications Standards Institute, 揇igitalVideo Broadcasting (DVB): Implementation guidelines for DVB terrestrial services; transmission aspects,?EuropeanTelecommunications Standards Institute, ETSI TR-101-190,1997. [Online]. Available: . [Accessed: Aug. 17, 1998].[27] J. Geralds, "Sega Ends Production of Dreamcast," ,para. 2, Jan. 31, 2001. [Online]. Available:/news/1116995. [Accessed Sept. 12,2004].[28] W. D. Scott & Co, Information Technology in Australia:Capacities and opportunities: A report to the Department of Science and Technology. [Microform]. W. D. Scott & Company Pty. Ltd. in association with Arthur D. Little Inc. Canberra: Department of Science and Technology, 1984.AbbreviationsStandard abbreviations may be used in your citations. A list of appropriate abbreviations can be found below:。
作文格式英语ppt课件
Use a readable font and an appropriate font size for the audience and presentation length
Writing the ending
Conclusion paragraph
The conclusion paragraph should summarize the main points and provide closure to the presentation
The use of punctation marks
Commas
Use commas to separate items in a list or to add emphasis or clarify information
Sessions
Use chromosomes to separate independent clauses or to separate items in a list with internal punctation
Variety
Vary the structure of your senses to make your writing more engaging and dynamic
The application of grammar
Standard English
Use standard English grammar and punctuation to ensure clarity
and consistency
Subordinate uses
Use suborder clauses to provide additional information or to connect ideas logically
英语写作Ⅱ-expository essays
Model Language and Culture
…… Once a group of Chinese were visiting the home of a fairly well-to-do American.As they were shown around the house, they commented, “you have a very nice home.It’s so beautiful.” The hostess smiled with obvious pleasure and replied in good American fashion “Thank you”—which caused surprise among some of her Chinese ter, while conversing at the dinner table, the host remarked to the Chinese interpreter, a young lady who had graduated not long ago from a Chinese university, “Your English is excellent.Really quite fluent.” To this she said, “No, no.My English is quite poor”—an answer that he had not expected and found a bit puzzling. Was the American hostess’ reply immodest, as it seemed to some of the Chinese? Was the young Chinese interpreter’s remark insincere, as it sounded to the Americans? In both cases the answer is no.……
基于卷积神经网络的古文字识别系统设计与实现
人工智能及识别技术本栏目责任编辑：唐一东基于卷积神经网络的古文字识别系统设计与实现陈盈祾，潘玉霞（三亚学院信息与智能工程学院，海南三亚572000）摘要：古文字作为中国上下五千年以来的使用文字，记录了我国从古至今的文化发展历史，对于我国的历史文化研究具有十分重要的作用。
对古文字的识别能够将那些珍贵的文献材料转换为电子文档，便于这些珍贵文献材料的保存和传播。
该文将深度学习中经典的卷积神经网络技术应用到古文字识别中，剖析了运用的卷积神经网络技术的原理结构，并阐述了系统在识别方面所运用的技术。
关键词：古文字识别；深度学习；卷积神经网络中图分类号：TP393文献标识码：A文章编号：1009-3044(2021)10-0207-02开放科学（资源服务）标识码（OSID ）：1前言古文字学——这一门古老但是却极其富有生命力的学科，在我们研究中国的古代历史以及文化中具有的十分重要的作用，它是打开古代历史文化宝库的一把钥匙。
我国历经上下五千年，文化厚重繁多，经过历史变迁，无数的朝代都拥有独属于自己的文化，尤其是文字。
文字最初的诞生传说是由于仓颉造字，后来随着历史演变，朝代更迭，文字慢慢进化。
在殷商时期有了我们熟悉的甲骨文，这是我们目前见到的最早的，较系统的成熟的文字。
再往后又进化出了金文，石鼓文，大小篆等。
目前，市面上现有的古文字识别系统可以根据用户所输入的简体汉字来查询出各个历史朝代对应的古文字。
但是，这些古文字识别系统却仅仅能够根据简体汉字来查询古文字，而不能通过古文字来查询简体字或者形近字。
而在考古方面出土文献的处理应用上，我们需要根据未知的古文字的字型来检视我们该文字已知的形近字及相关资料信息来辅助推断未知古文字的含义，例如：如果我们在某文物上发现刻有古文字，那我们如何快速的确定其是否为已知的古文字，抑或者我们又该如何快速地获取其已知形似字及该形近字的相关资料信息以便于我们推测其含义呢？如果没有一种技术或者产品能够辅助解决这个难题，无疑会给古文字工作者的工作带来极大的不便，影响古文字研究工作的迅速开展。