Understanding and Predicting Thiolated Gold Nanoclusters from First Principles
古代中国医学文献的命名实体识别研究——以Flat-lattice增强的SikuBERT预训练模型为例
*本文系国家社会科学基金项目“古代中国医学文献的知识标注与知识挖掘研究”（项目编号：17BTQ060）研究成果和国家社会科学基金重大项目“中国古代典籍跨语言知识库构建及应用研究”（项目编号：21&ZD331）研究成果。
古代中国医学文献的命名实体识别研究——以Flat-lattice 增强的SikuBERT 预训练模型为例*谢靖，刘江峰，王东波摘要标注古代中医文献的命名实体能挖掘其蕴藏的中医学知识，推进中医现代化发展。
文章基于BERT-base 、RoBERTa 、SikuBERT 、SikuRoBERTa 预训练模型，以《黄帝内经·素问》为研究对象、Flat-lattice Transformer （FLAT ）结构为微调模型，构建中医文献中病证、病理、经络、穴位、五行等命名实体识别任务。
实验结果表明：直接使用古文繁体BERT 模型对古代中医文献进行领域命名实体识别，基于繁体《四库全书》的SikuBERT 、SikuRoBERTa 预训练模型效果优于BERT-base 、RoBERTa 模型；引入FLAT 结构作为微调模型后，SikuBERT 在有标点情况下表现最优，识别效果可提升4%左右，SikuRoBERTa 在无标点情况下表现最优，识别效果可提高2%~3%。
实验验证了FLAT 作为微调模型对BERT 模型在中医专业领域中古文献命名实体识别工作上的有效性。
该微调模型可以有效避免分词错误引起的实体识别传播错误，提高中医命名实体的识别效率。
关键词命名实体识别中医典籍素问SikuBERT FLAT引用本文格式谢靖，刘江峰，王东波.古代中国医学文献的命名实体识别研究——以Flat-lattice 增强的SikuBERT 预训练模型为例[J].图书馆论坛，2022，42（10）：51-60.Study on Named Entity Recognition of Traditional Chinese Medicine Classics ：Taking SikuBERT Pre-training Model Enhanced by the Flat-lattice Transformer for ExampleXIE Jing ，LIU Jiangfeng &WANG DongboAbstractNamed Entity Recognition （NER ）can help to further excavate the knowledge of Traditional ChineseMedicine （TCM ），and promote the modernization of TCM as well.With the traditional Chinese Siku Quanshu version of Huangdi Neijin ·Suwen as the corpus ，this paper tries to testify the performance of BERT-base ，RoBERTa ，SikuBERT ，SikuRoBERTa pre-training models in NER of TCM classics ，especially in the recognition of named entities related to disease syndrome ，pathology ，meridian ，acupoint ，etc.The result shows that ：When directly used to perform NER of TCM classics ，the pre-training models of SikuBERT and SikuRoBERTa are better than those of BERT-base and RoBERTa.When enhanced and modified by the Flat-lattice Transformer （FLAT ），SikuBERT Pre-training Model performs best under the premise of punctuation ，and its recognition efficiency can be improved by about 4%；while SikuRoBERTa Pre-training Model performs best under the premise of no punctuation ，and its recognition efficiency can be improved by 2%~3%.In short ，FLAT performs well in the enhancement and modification of BERT pre-training models in NER of TCM classics ，and it should be further used to improve the recognition efficiency of TCM classics.Keywords named entity recognition ；traditional Chinese medicine classics ；Suwen ；SikuBERT ；FLAT0引言古代中国医学文献所记录的中医知识是中华传统科学文化的重要分支，对中医文献进行深度加工和知识标注，有助于挖掘蕴含在其中的古代医学知识及中医哲学思想。
ChatGPT在医学知识图谱构建中的应用：知识补充与结构化数据生成的智能助手(英文中文双语版优质文档
ChatGPT在医学知识图谱构建中的应用：知识补充与结构化数据生成的智能助手（英文中文双语版优质文档）The Application of ChatGPT in the Construction of Medical Knowledge Graph: An Intelligent Assistant for Knowledge Supplementation and Structured Data Generation (English and Chinese bilingual version high-quality documents)The medical knowledge graph is an important tool that can be used to integrate and structure knowledge in the medical field and provide intelligent assistants for healthcare. ChatGPT, as a powerful natural language processing model, can play a key role in the construction of medical knowledge graphs. This paper explores the application of ChatGPT in knowledge supplementation and structured data generation, as well as its potential as an intelligent assistant.1. Knowledge SupplementDuring the construction of medical knowledge graph, a key challenge is to obtain complete and accurate medical knowledge. Traditional methods require manual retrieval and integration of large volumes of literature and research, which is a time-consuming and labor-intensive task. However, ChatGPT can provide intelligent knowledge supplementation by processing natural language texts.ChatGPT can generate relevant answers and explanations based on given questions and background information. By training on medical literature and databases, ChatGPT can identify and extract medical concepts, disease symptoms, treatment methods and other information, and apply it to the construction of knowledge graphs. It can answer a variety of questions about diseases, drugs, diagnosis and treatment, helping doctors and researchers quickly obtain the medical knowledge they need.In addition, ChatGPT can also provide real-time research progress and the latest medical knowledge. By connecting with medical literature and academic databases, it can help users understand the latest medical research results and update the knowledge graph in a timely manner. This ability to supplement knowledge can help medical professionals stay informed of the latest scientific advances and provide more accurate and effective medical care.2. Structured data generationMedical knowledge graphs usually require structured data representation for association and query in the graph. However, converting unstructured medical text into structured data is a challenging task. At this time, ChatGPT can be used as a smart assistant to help automatically generate structured data.ChatGPT can extract important medical entities and relationships from natural language texts by understanding and parsing them. It can identify and label key information such as medical terms, disease names, drug dosages, etc., and convert them into a structured data format. The generation of such structured data can greatly improve the construction efficiency and quality of medical knowledge graphs.In addition, ChatGPT can also be combined with other natural language processing techniques, such as named entity recognition and relation extraction. By leveraging these techniques, ChatGPT can more accurately extract and organize medical knowledge and generate more structured data representations.in conclusionIn the construction of medical knowledge graphs, ChatGPT has important application potential as an intelligent assistant for knowledge supplementation and structured data generation. It can provide intelligent knowledge supplements by processing natural language texts to help medical professionals acquire comprehensive and accurate medical knowledge. At the same time, it can transform unstructured medical texts into structured data, and promote the construction and application of knowledge graphs. ChatGPT's powerful language understanding and generation capabilities make it an ideal partner in the medical field, which will promote the intelligent development of healthcare.医学知识图谱是一个重要的工具，可用于整合和结构化医学领域的知识，并为医疗保健提供智能助手。
基于ChatGPT的医学领域知识图谱问答系统：结合语义理解和推理的智能问答助手(英文中文版优质文档)
基于ChatGPT的医学领域知识图谱问答系统：结合语义理解和推理的智能问答助手（英文中文双语版优质文档）ChatGPT-based question answering system for knowledge graphs in the medical field: an intelligent question answering assistant combining semantic understanding and reasoning (high-quality documents in English and Chinese bilingual versions)Title: ChatGPT-Based Question Answering System for Medical Domain Knowledge Graph: An Intelligent Question Answering Assistant Combining Semantic Understanding and ReasoningPart I: IntroductionThe knowledge in the medical field is huge and complex. Researchers, doctors and patients need to obtain medical knowledge and answers quickly and accurately to support clinical decision-making and health management. Traditional information retrieval and intelligent question answering systems often perform poorly in the face of challenges in the medical field, because medical knowledge requires in-depth understanding and reasoning to get correct answers.The ChatGPT-based medical domain knowledge graph question answering system makes up for the shortcomings of traditional systems by combining semantic understanding and reasoning capabilities, and provides users with more intelligent and accurate question answering assistants. The system uses the ChatGPT model to train on a large-scale medical knowledge map, so that it has the ability to understand and answer questions in the medical field. This article will introduce the architecture and key technologies of the system in detail, as well as its application and performance evaluation on question answering tasks in the medical field.Part II: System Architecture and Key TechnologiesThe architecture of the knowledge graph question answering system in the medical field consists of three key components: a semantic understanding module, a reasoning module, and an answer generation module. The semantic understanding module is responsible for semantically analyzing the user's questions and converting them into queries on the knowledge graph. The reasoning module performs reasoning based on the knowledge and logic rules on the graph, helping the system understand and answer complex questions. The answer generation module is responsible for generating accurate and complete answers based on the reasoning results.In the semantic understanding module, we use a sequence-to-sequence model based on ChatGPT to convert questions into queries on graphs. The model can accurately capture the semantic information in the medical field by learning a large number of medical questions and corresponding queries, so as to better understand users' questions.The reasoning module makes use of entity relationships and logic rules in the medical knowledge graph for reasoning. Knowledge graphs contain rich medical entities (such as diseases, drugs, symptoms, etc.) and the relationships between them. The reasoning module can make use of these relationships to make inferences and inferences to solve complex problems raised by users. For example, when a user asks about the symptoms of a certain disease, the reasoning module can find the symptoms associated with the disease through the correlation in the knowledge graph, and return it to the user.The answer generation module generates accurate and complete answers according to the reasoning results and query sentences. According to the type and content of the question, it can choose the appropriate answer form, such as simple text answer, image or table, etc.Part III: Application and Performance EvaluationThe medical field knowledge map question answering system is widely used in the medical field. It can be used to help doctors and researchers find relevant medical literature and research results, and provide clinical guidelines and treatment recommendations. At the same time, patients and ordinary users can also obtain information on disease prevention and health management through the system.To evaluate the performance of the system, we conduct large-scale experiments and evaluations. We used standard question sets from several medical fields and invited professional physicians and researchers to participate in the assessment. The results show that the ChatGPT-based question answering system for knowledge graphs in the medical field is significantly better than traditional question answering systems in terms of accuracy and answer quality. It is able to correctly answer complex questions and provide detailed and useful information to meet users' needs.Part Four: Future Development DirectionsThere are still many directions that can be explored and improved for the ChatGPT-based knowledge graph question answering system in the medical field. First, we can further improve the accuracy and efficiency of the system in terms of semantic understanding and reasoning. Improve the system's ability to understand medical problems by optimizing the model's training algorithm and data processing technology.Second, we can extend the coverage and depth of the knowledge graph, enabling the system to answer more types of medical questions. This includes expanding the scope of medical entities and relationships, adding new knowledge and rules to meet the needs of the evolving medical field.In addition, we can also explore multi-modal question answering methods, combining multimedia information such as images and videos with knowledge graphs to provide richer and more intuitive answers.in conclusion:The ChatGPT-based medical field knowledge graph question answering system combines the capabilities of semantic understanding and reasoning to provide users with an intelligent and accurate question answering assistant. The system has great potential in the medical field to provide timely and reliable medical knowledge and answers to doctors, researchers and patients. Future research can further improve the system's accuracy and coverage to meet growing medical needs.标题：基于ChatGPT的医学领域知识图谱问答系统：结合语义理解和推理的智能问答助手第一部分：引言医学领域的知识庞大而复杂，研究人员、医生和患者需要快速准确地获取医学知识和答案以支持临床决策和健康管理。
笛卡尔的本体论之争
笛卡尔的本体论之争首先周一公布2001年6月18日;实质性修改太阳2006年10月15日笛卡尔的本体论（或先验）的论点，既是哲学的一个最迷人，他的理解方面的不足。
论据与魅力源于努力证明神的存在，从简单的处所，但功能强大。
存在是产生立即从清晰和明确的想法是一个无比完美。
讽刺的是，简单的说法也产生了一些误读，加剧了部分由笛卡尔没有一套单一版本。
该声明的论点主要出现在第五沉思。
这种说法因果来得早在接踵而至的一个神的存在，沉思在第三，不同的证据提出问题的两项之间的秩序和关系。
重复笛卡尔哲学原理，包括本体论争论的几个文本等中央。
他还辩解首先由一些主要的知识分子，他在一天，严厉打击反对第二次回复，和第五。
笛卡尔不是第一位哲学家，制订一个本体论的论点。
一个早期版本的说法已大力安瑟伦辩护圣在11世纪，然后圣托马斯阿奎那批评由当代），后来被命名为Gaunilo和尚（安瑟伦（尽管他的言论是针对然而，另一个版本参数）。
阿奎那的批评被视为如此具有破坏性，本体论的争论了数百年死亡。
它的出现，作为一个同时代的惊喜笛卡尔，他应该试图复活它。
虽然他声称没有被证明的熟悉安瑟伦的版本，笛卡尔似乎他自己的工艺参数，以阻止传统的反对。
尽管相似之处，笛卡尔的论点的版本不同于安瑟伦方式在重要的。
后者的版本被认为要从定义这个词的含义“上帝”，上帝是一个被一大于不能设想。
笛卡尔的观点相反，中，主要是基于两个他的哲学的中心原则-天生的思想理论和学说明确的印象和独特的。
他声称不依赖于上帝的任意定义，而是一种天生的想法，其内容是“的。
” 笛卡尔的版本也非常简单。
神的存在是直接从推断的事实，有必要存在的想法是包含在一个清晰而鲜明的超级完美的存在。
事实上，在一些场合，他建议，所谓的本体论“的论调”是不是一个正式的哲学偏见的证据，而是在所有不言而喻的公理直观地掌握了一个心灵的自由。
笛卡尔的本体论的争论相比往往以几何论证，认为有必要存在的想法不能排除再从神比事实平等的角度，其角度，例如两权，可以被排除在一个三角形的想法。
2024年6月第1套英语四级真题
大学英语四级考试2024年6月真题(第一套)Part I Writing (30 minutes) Directions: Suppose your university is seeking students'opinions on whether university libraries should be open to the public.You are now to write an essay to express your view.You will have 30 minutes for the task.You should write at least 120 words but no more than 180 words.Part ⅡListening Comprehension (25 minutes) Section ADirections: In this section,you will hear three news reports.At the end ofeach news report,you will hear two or three questions.Both the news report and the questions will bespoken only once.Afier you hear a question,you must choose the best answer from the four choices marked A),B),C)and D).Then mark the corresponding letter on Answer Sheet 1 with a single line through the centre.Questions 1 and 2 are based on the news report you have just heard.1.A)Due to a fire alarm in their apartments.B)Because of the smoke and heat damage2.A)Investigating the cause of the incident.B)Helping search for the suspect of the crime. C)Due to the water used to extinguish the flames.D)Becauseof the collapse of the three-story building.C)Rescuing the businessmen trapped in the building.D)Checking town records for the property developer.Questions 3 and 4 are based on the news report you have just heard.3.A)It plays a less important role in one's health than nutrient intake.B)It impacts people's health to a lesser degree than sun exposure.C)It is associated with people's mental health conditionsD)It is linked with older adults'symptoms ofdepression4.A)It was indefinite C)It was straightforward.B)It was systematic. D)It was insignificant. Questions 5 to 7 are based on the news report you have just heard.5.A)It has helped solve several murder cases.B)It has become a star police dog in Beijing6.A)To speed up investigation into criminal cases.B)To test the feasibility of cloning technology.7.A)Cloning is too complicated a processB)The technology is yet to be accepted C)It has surpassed its mother in performance.D)It has done better than naturally born dogs.C)To cut down training expensesD)To reduce their training time.C)Cloning is ethically controversial.D)The technology is too expensive.Section BDirections: In this section,you will hear two long conversations.At the end ofeach conversation,you will hearfour questions.Both the conversation and the questions will bespoken only once.After you hear a question,you must choose the best answer from the four choices marked A),B),C)and D).Then mark the corresponding letter on Answer Sheet I with a single line through the centre.Questions 8 to 11 are based on the conversation you have just heard.8.A)He read it somewhere online.B)He heard about it from a coworker.9.A)His publications.B)His first book.10.A)Collect a lot more data.B)Relax a bit less often.11.A)Find out the show's most interesting episodesB)Watch the series together with the woman. C)He read an article reviewing it.D)He watched a TV series based on it.C)His addressD)His name.C)Clarify many new conceptsD)Read more reference books.C)Get an e-Copy of the book to read.D)Check to see when the show starts.Questions 12 to 15 are based on the conversation you have just heard.12.A)To check the prices of his farm produce.B)To ask the way to the Newcastle City Hall.13.A)Bakers.B)Vendors14.A)The issuing of certificates to vendors.B)The completion of the baking task.15.A)The closing date of submission. C)To inquire about the vegetarian food festival.D)To seekthe man's help with her work on the farm.C)Vegetarians.D)OrganisersC)The festival they are organising.D)The deadline for application.C)The details of the ceremonyB)The website of his company. D)The organiser'saddressSection CDirections: In this section,you will hearthre passages.At the end ofeach passage,you will hear three or four questions.Both the passage and the questions will be spoken only once.Afteryou hear a question,you must choose the bestanswer from the four choices marked A),B),C)and D).Then mark the corresponding letter on Answer Sheet I with a single line through the centreQuestions 16 to 18 are based on the passage you havejust heard.16.A)Most scenic sites have been closed.B)Access to official campsites is limited17.A)It is strongly opposed by nearby residentsB)It leads to much waste of public money18.A)Look for open land in ScotlandB)Leave no trace of their camping C)Health experts advise going outdoors.D)People have more time during the summer.C)It has caused environmental concernsD)It has created conflicts among campers.C)Avoid getting close to wilderness.D)Ask for permission from authorities.Questions 19 to 21 are based on the passage you have just heard.19.A)They outcompete mythical creatures.B)They usually mind their own business. C)They truly exist in the AmazonregionD)They resemble alarmingly large snakes20.A)Scar tissue from dolphins'fighting.B)Skin infection from water pollution.21.A)It has been shrinking at an astonishing pace. C)Unhealed wounds from snake bites.D)Swimming along in seasonal floods.B)It has been placed under international protection.C)It has been appealing to both freshwater and sea dolphinsD)It has been abandoned as a battleground for male dolphins.Questions 22 to 25 are based on the passageyou have just heard.22.A)About 58%of young adults call parental support the new normal.B)Most adult children enjoy increasing sources of financial supportC)A full 70%of the young adults cannot afford to buy a car by themselves.D)Most early adults cannot sustain their lifestyles without parental support23.A)It renders them dependent. C)It makes them mentally immature.B)It causes them to lose dignity. D)It hinders them from getting ahead.24.A)It challenges one's willpower C)It calls for due assistance.B)It results from education. D)It defines adulthood.25.A)Current lifestyles C)College loansB)Poor budgeting. D)Emergency expensesPart ⅢReading Comprehension (40 minutes) Section ADirections: In this section,there is a passage with ten blanks.You are required to select one word for each blank from a list of choices given in a word bank following the passage.Read the passage through carefully before making your choices.Each choice in the bank is identified by a letter:Please mark the corresponding letter for each item on Answer Sheet 2 with a single line through the centre.You may not use any of the words in the bank more than once.It's well known that physical exercise is beneficial not just to physical health but also to mental health.Yet whereas most countries have 26 evidence-backed guidelines on the type and intensity of exercise 27 for various physical health benefits,such guidelines do not yet exist for exercise and mood. This is 28 due to a lack of necessary evidence.However,a new systematic review brings us usefully up- to-date on the current findings in this area.Before 29 into some of the key take-aways,an important 30 made in the review is between aerobic exercise and anaerobic.The former 31 such things as walking,jogging and cycling and meansexercising in such a way that your body is able to use oxygen to burn fat for energy.In contrast,anaerobic exercise—such as lifting heavy weights—is of such 32 intensity that your body does not have time to use oxygen to create energy and so instead it breaksdown glucose(葡萄糖)in your blood or muscles.Beginning first with the influence of exercise intensity on the mood benefits of aerobic exercise,the researchers,led by John Chan at Shenzhen University,found 33 resultsfrom 19 relevant studies.Somefavoured higher intensity,others low,while seven studies found that intensity made no 34 _to mood benefits.In relation to the intensity of anaerobic exercise,however,the results were far clearer—the optimum (最佳选择)for improving mood is 35 intensity,perhaps because low intensity is too dull while high intensity is too unpleasantSection BDirections: In this section,you are going to read a passage with ten statements attached to it.Each statement contains information given in one of the paragraphs.Identify the paragraph from which the information is derived.You may choose a paragraph more than once.Each paragraph is marked with a letter.Answer the questions by markingthe corresponding letter on Answer Sheet 2.Why DoAmericans Work So Much?A)How will we all keep busy when we only have to work 15 hours a week?That was the question that worriedthe British economist John Maynard Keynes when he wrote his short essay“Economic Possibilities for Our Grandchildren”in1930.Over the next century,he predicted,the economy would become so productive that people would barely need to work at all.For a while,it looked like Keynes was right.In 1930 the average working week was 47 hours in the United States.But by 1970,the number of hours Americans worked on average had fallen to slightly less than 39.B)But then something changed.Instead of continuing to decline,the duration of the working weekremained stable.It has stayed at just below 40 hours for nearly five decades.So what happened?Why are people working just as much today as in 1970?C)There would be no mystery in this if Keynes had been wrong about the power of technology to increase theeconomy's productivity,which he thought would lead to a standard of living “between four and eight times as high as it is today.”But Keynes got that right:Technology has made the ec onomy massively moreproductive. According to Benjamin M.Friedman,an economistat Harvard,the U.S.economy is right ontrack to reach Keynes's eight-fold(八倍)multiple by2029.That is a century after the last data Keynes wouldhave had access to.D)In a new paper,Friedman tries to figure out why that increased productivity has not translated into increasedleisure time.Perhaps people just never feel materially satisfied,always wanting more money to buy the nextnew thing.This is a theory that appeals to many economists.“This argument is,at best,far from sufficient,”he writes.If that were the case,why did the duration of the working week decline in the first place?E)Another theory Friedman considers is that,in an era of ever fewer settings that provide effectiveopportunitiesfor personal connections and relationships,people may place more value on the socializing that happens at work.There is support for this theory.Many people today consider colleagues as friends.But Friedman argues that the evidence for this theory is far from conclusive.Many workers report that they would like to spend more time with family,rather than at work.Furthermore,this theory cannot explain the change in trend in the U.S.working week in the 1970s.F)A third possibility proves more convincing for Friedman.That is:American inequality means that the gainsof increasing productivity are not widely shared by everyone.In other words,most Americans are too poor to work less.Unlike the other two explanations Friedman considers,this one fits chronologically(按年代).Inequality declined in America during the period following World War II,along with the duration of the working week.But since the early 1970s it has risen dramatically.G)Keynes's prediction of a shorter working week rests on the idea that the standard of living would continuerising for everyone.But Friedman says that this is not what has happened.Although Keynes's eight-fold figure holds up for the economy as a whole,it is not at all the case for the median(中位数的)American worker.For them,output by 2029 is likely to be around 3.5 times what it was when Keynes was writing.This is a bit below his four-to eight-fold predicted rangeH)This can be seen in the median worker's income over this time period,complete with a shift in 1973 that fitsin precisely with when the working week stopped shrinking.According to Friedman,between 1947 and 1973 the average hourly wage for normal workers (those who were not in management roles)in private industries other than agriculture nearly doubled in terms of what their money could buy.But by 2013 the average hourly wage for ordinary workers had fallen 5 percent from the 1973 level in terms of actual purchasing power.Thus,though American incomes may have gone up since 1973,the amount that American workers can actually buy with their money has gone down.For most Americans,then,the magic of increasing productivity stopped working around 1973.Thus,they had to keep working just as much in order to maintain their standard of livingI)What Keynes predicted was a very optimistic version of what economists call technological unemployment.This is the idea that less labor will be necessary because machines can do somuch.In Keynes's vision,the resulting unemployment would be distributed more or less evenly across society in the form of increased leisure.But Friedman says that,for Americans,reality is much darker.Americans now have a labor market in which millions of people—those with fewer skills and less education—are seeking whatever poorly paid work they can get.This is confirmed by a recent poll that found that,for half of hourly workers,their top concern is not that they work too much but that they work too little.This is most likely not because they like their jobs so much.Rather,we can assume it is because they need the money.J)This explanation leaves an important question.If the very rich—the workers who have reaped above-average gains from the increased productivity since Keynes's time—can afford to work less,why do they continue to work so much?(Indeed,research has shown that the highest earners in America tend to work the most.)Friedman believes that for many top earners,work is a labor of love.They are doing work they care about and are interested in,and doing more of it is not necessarily a burden.For them,it may even be a pleasure.Thesetop earners derive meaning from their jobs and work is an important part of how they think of themselves.And,of course,they are compensated for it at a level that makes it worth their while.K)Friedman concludes that the prosperity(繁荣)Keynes predicted is here.After all,the economy as a whole has grown even more brilliantly than he expected.But for most Americans,that prosperity is nowhere to beseen.And,as a result,neither are those shorter working weeks.36.Some people view socializing at the workplace as a chance to develop personal relationships.37.As ordinary American workers'average hourly pay had decreased despite increasing productivity,they hadto work just as manyhours as before to keep their living standards.38.American workers'average weekly workingtime has not changed for nearly half a century.39.Friedman believes inequality in the rgely explains why increasing productivity has not resulted inreduced working hours.40.Many economists assume people's thirst for material things has prevented them from enjoying moreleisure time.41.An economist'sprediction about a shorter average working week seemed to be correct for a time in the 20thcentury.42.In the bor market,the primary concern of people with less schooling and fewer skills is to secure anyemployment even if it is low-paid.43.Keynes was right in predicting that technology would make the economy much more productive.44.Many of the highest earners have a keen interest in and love for what they are doing45.According to Keynes,there would be a shorter working week with everyone's standard of living continuingtorise.Section CDirections:There are 2 passages in this section.Eachpassage is followed by some questions or unfinished statements.For each of them there are four choices marked A),B),C)and D).You should decide on the best choice and mark the corresponding letter on Answer Sheet 2 with a single line through the centre.Passage OneQuestions 46 to 50 are based on the following passage.Lao Zi once said,“Care about what other people think and you will always be thei r prisoner.”People-pleasing,or seeking self-worth through others'approval,is unproductive and an exhausting way to go through life.Why do we allow what others think of us to have so much power over how we feel about ourselves?If it's true that you can't please all people all of the time,wouldn't it make sense to stop trying?Unfortunately,sense often isn't driving our behavior.For social beings who desire love and belonging, wanting to be liked,and caring about the effect we have on others,is healthy and allows us to make connections.However,where we get into trouble is when our self-worth is dependent upon whether we win someone's approval or not.This need to be liked can be traced back to when we were children and werecompletely dependent on others to take care of us:Small children are not just learning how to walk and communicate,they are alsotrying to learn how the world works.We learn about who we are and what is expected of us based on interactions with others so,to a four-year-old,if Mommy or Daddy doesn't like him or her,there is the danger that they will abandon them.We need to understand that when we desperately want someone to approveof us,it's being driven by thatlittle kid part of us that is still terrified of abandonment.As you become more capable of providing yourself with the approval you seek,your need for external validation will start to vanish,leaving you stronger,more confident,and yes,happier in your life.Imagine howmuch time we lose each moment we restrainour authentic selves in an effort to be liked.If we base our worth on the opinions of others,we cheat ourselves of the power to shape our experiences and embrace life not only for others but also for ourselves,becauseultimately,there is no difference.So embrace the cliché(老话)and loveyourself as it's highly doubtful that you'll regret it.46.What can we conclude from Lao Zi's quotation?A)We should seethrough otherpeople's attempt to make a prisonerof us.B)We can never really please other people even if we try as hard as we can.C)We can never be truly free if taking to heart others'opinion of us.D)We should care about other people's view as much as they care about our own.47.What will happen if we base our self-worth on other people's approval?A)Our desire to be loved will be fulfilled. C)Our identity as social beings will be affected.B)Our life will be unfruitful and exhausting D)Our sense of self will be sharpened and enhanced.48.What may account for our need to be liked or approved of?A)Our desperate longing for interactions with others.C)Our knowledge about the pain of abandonment.B)Our understanding of the workings of the world. D)Our early childhood fear of being deserted.49.What can we do when we become better able to provide ourselves with the desired approval?A)Enjoy a happier life.C)Receive more external validation.B)Exercise self-restraint.D)Strengthen our power of imagination.50.What does the author advise us to do in the last paragraph?A)Embrace life for ourselves and for others. C)See our experiencesas assets.B)Base our worth on others'opinions. D)Love ourselves as we arePassage TwoQuestions 51 to 55 are based on the following passage.Some people have said aging is more a slide into forgetfulness than a journey towards wisdom.However,a growing body of research suggests that late-in-life learning is possible.In reality,education does an aging brain good.Throughout life,people's brains constantly renovate themselves.In the late 1960s,British brain scientist Geoffrey Raisman spied growth in damaged brain regions ofrats through an electron microscope; their brains were forging new connections.This meant brains may change every time a person learns something new.Of course,that doesn't mean the brain isn't affected by the effects of time.Just as height usually declines over the years,so does brain volume:Humans lose about 4 percent every decade starting in their 40s.But that reduction doesn't necessarily make people think slower;as long as we are alive and functioning,we can alter our brains with new information and experiences.In fact,scientists now suspect accumulating novel experiences,facts,and skills can keep people's minds more flexible.New pathways can strengthen our ever-changing mental structure,even as the brain shrinks.Conventional fixes like word puzzles and brain-training apps can contribute to mental durability.Even something as simple as taking a different route to the grocery store or going somewhere new on vacation can keep the brain healthy.A desire for new life challenges can further boost brainpower.Research about aging adults who take on new enterprises shows improved function and memory as well as a reduced risk of mental disease.Openness—a characteristic defined by curiosity and a desire for knowledge—may also help folks pass brain tests.Some folks are born with this take-in-the-world atitude,but those who aren't as genetically gifted aren't necessarily out of luck.While genes can encourage an interest in doing new things,a 2012 study in the journal Psychology and Aging found completing reasoning tasks like puzzles and number games can enhance that desire for novel experiences,which can,in turn,refresh the brain.That's why brain scientist Richard Kennedy says “It's not that old dogs can't learn new tricks.It's that maybe old dogs don't realize why they should.”51.What do some people think of aging adults?A)Their wisdom grows as time goes by. C)They can benefit from late-in-life learning.B)Their memory gradually deteriorates D)They are likely to have mental health issues.52.What can we conclude from Geoffrey Raisman's finding?A)Brain damage seriously hinders one'slearning. C)Brains can refresh and improve with learning.B)Brain power weakens slower than we imagine D)Brains forge connections under new conditions53.What is one thing that helps maintain the health of our brain even as it shrinks?A)Doing daily routines by conventional means. C)Imitating old dogs'way of learning new tricksB)Avoiding worrying about our mental durability. D)Approaching everyday tasks in novel ways.54.What does the author say can contribute to the improvement of brain function?A)Being curious and desiring knowledge. C)Rising to life's challenges and avoiding risks.B)Being eager to pass brain tests at an old age. D)Boosting immunity to serious mental diseases55.What is the finding of the 2012 study in the journal Psychology and Aging?A)Wishing to solve puzzles enhances one's reasoning power.B)Playingnumber games unexpectedly stimulates one's memory.C)Desiring new experiences can help to renovate thebrain.D)Learning new tricks shouldnot beconfined to old dogs only.Part IV Translation (30 minutes) Directions: For this part,you are allowed 30 minutes to translate a passage from Chinese into English.You should write your answer on AnswerSheet 2.四合院(siheyuan) 是中国一种传统的住宅建筑，其特点是房屋建造在一个院子的四周，将院子合围在中间。
conversation用法总结
Conversation用法总结1. 概述Conversation是一种人与机器之间进行对话的方式，它允许用户提出问题或发表陈述，并从机器中获取有关特定主题的信息。
在人工智能领域，Conversation被广泛应用于各种任务，如聊天机器人、智能助手和客服系统等。
通过理解和生成自然语言，Conversation使得机器能够模拟人类对话，为用户提供个性化的服务和支持。
2. Conversation的重要观点2.1 自然语言理解（Natural Language Understanding, NLU）自然语言理解是Conversation中的重要环节，它涉及将用户输入的自然语言文本转换为可理解和处理的形式。
NLU技术通常包括词法分析、句法分析、语义分析等子任务，旨在从文本中提取出关键信息，并确定用户意图和上下文。
2.2 对话管理（Dialog Management）对话管理是Conversation中的关键组成部分，它负责根据用户输入和系统状态来决定如何生成回复。
对话管理涉及到对上下文进行建模和维护，以便能够正确地响应用户，并采取适当的行动。
常用的对话管理方法包括基于规则、基于有限状态机和基于强化学习的方法。
2.3 自然语言生成（Natural Language Generation, NLG）自然语言生成是Conversation中的另一个重要环节，它负责将机器生成的信息转换为自然语言文本，以便向用户传达回复。
NLG技术通常涉及到文本生成、语音合成等任务，旨在产生流畅、连贯且符合语法规则的输出。
2.4 多轮对话（Multi-turn Conversation）多轮对话是Conversation中常见的场景之一，它涉及到用户和机器之间进行多次交互来完成一个任务。
在多轮对话中，对话管理起着至关重要的作用，需要能够正确地理解上下文、处理用户意图并生成合适的回复。
2.5 评估与优化（Evaluation and Optimization）评估与优化是Conversation系统开发过程中必不可少的一环。
2024上海松江区高三英语二模
松江区2023学年度第二学期模拟考质量监控试卷高三英语（满分140分，完卷时间120分钟）2024.4 考生注意：1．本考试设试卷和答题纸两部分，试卷包括试题与答题要求，所有答题必须涂（选择题）或写（非选择题）在答题纸上，做在试卷上一律不得分。
2．答题前，务必在答题纸上填写学校、班级、姓名和考号。
3．答题纸与试卷在试题编号上是一一对应的，答题时应特别注意，不能错位。
Ⅰ．Listening ComprehensionSection ADirections: In Section A, you will hear ten short conversations between two speakers. At the end of each conversation, a question will be asked about what was said. The conversations and the questions will be spoken only once. After you hear a conversation and the question about it, read the four possible answers on your paper, and decide which one is the best answer to the question you have heard.1．A．At 8:00. B．At 8:15. C．At 8:30. D．At 8:45.2．A．A professor. B．A coach. C．An engineer. D．A nurse.3．A．In a restaurant. B．In a hairdres ser’s.C．At a cinema. D．At a tailor’s. 4．A．Ways to visit a university. B．Two student tour guides.C．A tour of Fudan University. D．The campus of Fudan University.5．A．They did not make it there finally.B．They were not well received there.C．They experienced something unpleasant on the way.D．They had a wonderful time before they arrived there.6．A．Excited. B．Interested. C．Confused. D．Annoyed. 7．A．Practice the presentation in front of him. B．Watch how he makes a presentation. C．Reduce the time spent in practicing. D．Find out who her audience will be.8．A．She is always absent-minded. B．She forgot to tell the man about it.C．She is unclear about Sophie’s plan.D．She slipped in the neighboring town. 9．A．Because it took him much time to go to work.B．Because he had to save money for his journey.C．Because the job arranged many business journeys.D．Because he considered it unlucky to have that job.10．A．Buy a new printer with less noise. B．Ask the man to borrow a printer.C．Read a book on how to fix the printer. D．Get someone to repair the printer.Section BDirections: In Section B, you will hear two passages and one longer conversation. After each passage or conversation, you will be asked several questions. The passages and the conversation will be read twice, but the questions will be spoken only once. When you hear a question, read the four possible answers on your paper, and decide which one is the best answer to the question you have heard.Questions 11 through 13 are based on the following passage.11．A．How encores came into existence. B．How bands perform encores properly.C．Why audiences used to need encores. D．Why encores are part of a performance. 12．A．The 17th century. B．The 18th century. C．The 19th century. D．The 20th century. 13．A．French people were more interested in encores than others.B．Bands usually prepare more than two encores for each show.C．Recording technology boosted audiences’ needs for encores.D．Musicians can get recharged during the break before encores.Questions 14 through 16 are based on the following passage.14．A．Because of the rule for the class. B．Because of the course materials.C．Because the speaker changed his topics. D．Because the speaker disliked technology. 15．A．The students do not assess the speaker’s class fairly.B．The students are satisfied with the class environment.C．The speaker did not favor leaving technology at the door.D．The speaker were worried about students’ evaluation on him.16．A．It will stop students getting on well together.B．It may help students better understand themes.C．It will improve teaching effect by giving students more help.D．It may distract students from digging deep within themselves.Questions 17 through 20 are based on the following conversation.17．A．Doctor and patient. B．Salesman and customer.C．Teacher and student. D．Employer and employee.18．A．Fishing industry. B．Statistics. C．Computer modeling. D．Note-taking. 19．A．She is good at making model computers. B．She has decided on the title of the essay. C．She is uninterested in coping with statistics. D．She has always been weak at note-taking. 20．A．Learn to take notes immediately. B．Find out possible strategies alone.C．Read for more useful information. D．Work on her weaknesses by herself. Ⅱ．Grammar and VocabularySection ADirections: After reading the passage below, fill in the blanks to make the passage coherent and grammatically correct. For the blanks with a given word, fill in each blank with the proper form of the given word; for the other blanks, use one word that best fits each blank.Remote Work Slows Senior Housing Market RecoveryWith the rise of remote work, the market for senior housing has met with problems in its recovery. Only a few old people choose to live in senior-living communities （21）______the growing senior population and the cancelation of COVID-19 restrictions once making family visits difficult.（22）______ this trend suggests is that people’s shift to remote work contributes to the slow rebound of the senior housing market. That is, remote work is keeping many older Americans from moving into senior-living communities once warmly （23）______（welcome）.When more adults began working remotely during the pandemic（流行病）, they were able to check in on aging parents easily —they （24）______ take care of their parents’ issues on short notice.Experts have been analyzing the phenomenon in different ways. Some found that the greater flexibility to care for parents （25）______（mean）people’s delay in sending aged parents to expensive senior-housing accommodations. Therefore, markets with high levels of people working from home usually have lower senior-housing occupancy rates. Others said remote work might have some effect but also pointed to different factors. For instance, many seniors think that their family wallets are getting thinner, making some of them reluctant （26）______（send）to senior-living communities.The age at which people enter senior housing is also increasing, （27）______serves as another sign that shows people are choosing to delay transitioning. The rising cost of senior living weighs heavily on that decision. The CPI （consumer-price index）for nursing homes and adult day services rose 4.5% last May compared with （28）______in May, 2022.Still, many senior-housing operators are optimistic. When （29）______（illustrate）their point, they showed an increase in the number of people turning 80 years old over the following years and the actual wealth they have collected. Moreover, they find remote work arrangements are decreasing in some parts of the country, （30）______ employees there have seen their lowered productivity while working from home.Section BDirections: After reading the passage below, fill in each blank with a proper word chosen from the box. Each word can be used only once. Note that there is one word more than you need.A．accompanied B．allowed C．feasibly D．fueled E．intensity F．option G．prompting H．routine I．surgically J．underlying K．variedBrain Signals for Lasting PainBrain signals that reveal how much pain a person is in have been discovered by scientists who say the work is a step towards new treatments for people living with lasting pain.It is the first time researchers have decoded the brain activity 31 patients’ lasting pain. That has raised the hope that brain stimulation treatment alre ady used for Parkinson’s and major depression can help those running out of any other 32 . “We’ve learned that lasting pain can be tracked and predicted in the real world,” said Prasad Shirvalkar, lead researcher on the project at the University of California.Lasting pain affects nearly 28 million adults in the UK alone, and the causes are 33 . ranging from cancer to back problems. That being the case, lasting pain has 34 a rise in taking powerful painkillers. But nomedical treatments work well for the condition, 35 experts to call for a complete rethink in how health services handle patients with lasting pain.For the latest study, Shirvalkar and his colleagues 36 implanted electrodes（电极）into four patients with lasting pain hard to deal with after the loss of legs. The devices 37 the patients to record activity and collect data in two brain regions—the ACC and the OFC—at the press of one button on a remote handset. Several times a day, the volunteers were asked to complete short surveys on the 38 of pain, meaning how strong the pain was, and then record their brain activity. These scientists, armed with the survey responses and brain recordings, found they could use computers to predict a person’s pa in based on the electrical signals in their OFC. “We found very different brain activity 39 severe pain and have developed an objective biomarker for that kind of pain,” said Shirvalkar. The finding may explain, at least in part, why 40 painkillers are less effective for lasting pain. “The hope is that we can use the information to develop personalized brain stimulation treatment for the most severe forms of pain.”Ⅲ．Reading ComprehensionSection ADirections: For each blank in the following passage there are four words or phrases marked A, B, C and D. Fill in each blank with the word or phrase that best fits the context.The way of recording things has never ceased to develop. In the 1980s, as sales of video recorders went up, old 8mm home movies were gradually replaced by VHS （video home system）tapes. Later, video tapes of family holidays lost their appeal and the use of DVDs 41 . Those, too, have had their day. Even those holding their childhood memories in digital files on their laptops now know these files face the risk of 42 .Digitising historical documents brings huge benefits—files can be 43 and distributed, reducing the risk of their entire loss through physical damage caused by fire or flooding. And developing digital versions reduces44 on the original items. The International Dunhuang Project, 45 , has digitised items like manuscripts（手稿）from the Mogao caves in China, enabling scholars from around the world to access records easily without touching the real items.But the news that the Ministry of Justice of the UK is proposing to scan the 110 million people’s wills it holds and then destroy a handful of 46 after 25 years has shocked historians. The ministry cites this as a way of providing easier access for researchers. But that only justifies digitisation, not the 47 of the paper copies. The officials note the change will be economically efficient （saving around £4.5m a year）while keeping all the essential information.Scholars 48 . Most significantly, physical records can themselves carry important information — the kind of ink or paper used may be part of the history that historians are 49 . and error s are often made in scanning. Besides, digital copies are arguably more 50 than the material items, just in different ways. The attack from the Internet on the British Library last October has prevented scholars from 51 digitised materials it holds: imagine if researchers could not return to the originals. Some even think digitised information can easily be lost within decades no matter what 52 are put in place.The government says that it will save the original wills of “famous people for historic record”, such as that of Princess Diana’s. However, assuming that we know who will 53 to future generations is extraordinarilyproud. Mary Seacole, a pioneering nurse who now appears on the national school course in the UK, was largely54 for almost a century.The digitisation of old documents is a valuable, even essential measure. But to destroy the originals once they have been scanned, is not a matter of great 55 , but of huge damage.41．A．paused B．boomed C．recovered D．disappeared 42．A．getting outdated B．coming into style C．being fined D．making an error 43．A．deleted B．named C．copied D．altered 44．A．fight or flight B．life or death C．wear and tear D．awe and wonder 45．A．unfortunately B．additionally C．in summary D．for example 46．A．the originals B．the essentials C．the visualised D．the digitised 47．A．preservation B．classification C．publication D．destruction 48．A．applaud B．disagree C．discriminate D．withdraw 49．A．revising B．abandoning C．uncovering D．enduring 50．A．meaningful B．favourable C．resistant D．delicate 51．A．inventing B．adjusting C．accessing D．damaging 52．A．outcomes B．safeguards C．deadlines D．byproducts 53．A．matter B．respond C．lose D．live 54．A．spared B．discussed C．forgotten D．protected 55．A．sacrifice B．courage C．efficiency D．admirationSection BDirections: Read the following three passages. Each passage is followed by several questions or unfinished statements. For each of them there are four choices marked A, B, C and D. Choose the one that fits best according to the information given in the passage you have just read.（A）Charles Robert Darwin was born on 12 February 1809 in Shropshire, England. Darwin’s childhood passion was science, and his interest in chemistry, however, was clear; he was even nicknamed ‘Gas’ by his classmates.In 1825, his father sent him to study medicine at Edinburgh University, where he learned how to classify plants. Darwin became passionate about natural history and this became his focus while he studied at Cambridge. Darwin went on a voyage together with Robert Fitzroy, the captain of HMS Beagle, to South America to facilitate British trade in Patagonia. The journey was life-changing. Darwin spent much of the trip on land collecting samples of plants, animals and rocks, which helped him to develop an understanding of the processes that shape the Earth’s surface. Darwin’s analysis of the plants and animals that he gathered led him to express doubts on former explanations about how species formed and evolved over time.Darwin’s work convinced him that natural selection was key to understanding the development of the natural world. The theory of natural selection says that individuals of a species are more likely to survive when they inherit （经遗传获得）characteristics best suited for that specific environment. These features then become more widespread and can lead eventually to the development of a new species. With natural selection, Darwin argued how a wide variety of life forms developed over time from a single common ancestor.Darwin married his cousin, Emma Wedgwood, in 1839. When Darwin’s eldest daughter, Annie, died from a sudden illness in 1851, he lost his belief in God. His tenth and final child, Charles Waring Darwin, was born in 1856．Significantly for Darwin, this baby was disabled, altering how Darwin thought about the human species. Darwin had previously thought that species remained adapted until the environment changed; he now believed that every new variation was imperfect and that a struggle to survive was what drove species to adapt.Though rejected at the beginning, Darwin’s theory of evolution by natural selection is nowadays well acc epted by the scientific community as the best evidence-based explanation for the diversity and complexity of life on Earth. The Natural History Museum’s library alone has 478 editions of his On the Origin of Species in 38 languages.56．What made Darwin reconsider the origin and development of species?A．Examining plants and animals collected.B．His desire for a voyage to different continents.C．Classifying samples in a journey to South America.D．His passion for natural history at Edinburgh University.57．We can learn from paragraphs 1 to 3 that Darwin ______.A．used natural selection to develop new speciesB．enjoyed being called nicknames related to scienceC．learned some knowledge about plants when studying medicineD．argued with others over the diversity of life forms for a long period58．Which of the following changed Darwin’s view on the human species?A．That he had ten children in all. B．His youngest son’s being disabled.C．That he lost his eldest daughter. D．His marriage with Emma Wedgwood.59．This passage is mainly about ______.A．Darwin’s passion for medical science B．Darwin’s theory and experimentsC．Charles Darwin’s changing interest D．Charles Darwin’s life and work（B）Welcome to Muir Woods! This rare ancient forest is a kingdom of coast redwoods, many over 600 years old. How to get here?People using personal vehicles must have reservations before arriving at the park. （Details at.）Muir Woods National Monument is open daily, 8 a. m. to sunset. Stop by Visitor Center to get trails（路线）and program information, and to take in exhibits.What’s your path?Enjoy a walk on the paved Redwood Creek Trail （also called Main Trail）. Choose short, medium, or long loops（环线）. Other trails go deep into Muir Woods and Mount Tamalpais State Park.（Refer to the map of Muir Woods on the right for details.）Ready to explore more?Muir Woods is part of Golden Gate National Recreation Area, which includes Marin Headlands, Alcatraz, the Presidio, and Ocean Beach. Download the app at /goga.Stay safe and protect your park.Wi-Fi and cell service are not available. ·Watch for poisonous plants and falling branches. ·Do not feed or disturb animals. ·Fishing is prohibited in the park. ·Do not mark or remove trees, flowers, or other natural features. ·Go to the park website for more safety tips and regulations.AccessibilityWe make a great effort to make facilities, services, and programs accessible to all. For information, go to Visitor Center, ask a ranger, call, or check our website.More InformationMuir Woods National Monument /muwo Mill Valley, CA 94941-269660．Muir Woods will probably attract ______.①redwood lovers ②hunting lovers ③fishing lovers ④hiking loversA．①②B．③④C．①④D．②③61．What can be learned from the passage?A．Muir woods is surrounded by highland and ocean beaches.B．Visitors can read electronic maps using Wi-Fi in Muir Woods.C．Visitors are advised to call Visitor Center for safety tips and regulations.D．Reservations should be made if visitors drive private cars to Muir Woods.62．According to the map of Muir Woods, ______.A．Bridge 4 is the farthest from the parking lots of all bridgesB．Mill Valley is located on the southwest side of Muir BeachC．Bootjack Trail can lead one to Visitor Center from Bridge 3D．food and gifts can be bought on various sites in Muir Woods（C）Precognitive dreams are dreams that seemingly predict the future which cannot be inferred from actually available information. Former US President Abraham Lincoln once revealed the frightening dream to his law partner and friend Ward Hill Lamon, “…Then I heard people weep… ‘Who is dead in the White House?’ I demanded. ‘The President,’ ‘he was killed!’…” The killing did happen later.Christopher French, Professor in the Department of Psychology at Goldsmiths, stated the most likely explanation for such a phenomenon was coincidence（巧合）. “In addition to pure coincidences we must also consider the unreliability of memory”, he added. Asked what criteria would have to be met for him to accept that precognitive dreams were a reality, he said, “The primary problem with tests of the claim is that the subjects are unable to tell when the event（s）they’ve dreamed about will happen.”However, some claimed to make such tests practicable. Professor Caroline Watt at the University of Edinburgh, has conducted studies into precognitive dreaming. She stated that knowing future through dreams challenged the basic assumption of science — causality （relationship of cause and effect）.Dick Bierman, a retired physicist and psychologist, who has worked at the Universities of Amsterdam, Utrecht and Groningen, has put forward a theory that may explain precognitive dreams. It is based on the fact that when scientists use certain mathematical descriptions to talk about things like electromagnetism（电磁学）, these descriptions favour the belief that time only moves in one direction. However, in practice the wave that is running backwards in time does exist. This concept is called the time symmetry, meaning that the laws of physics look the same when time runs forward or backward. But he believes that time symmetry breaks down due to external conditions. “The key of the theory is that it assumes that there is a special context that restores th e broken time-symmetry, if the waves running backwards are ‘absorbed’ by a consistent multi-particle（多粒子）system. The brain under a dream state may be such a system where broken time-symmetry is partially restored. This is still not a full explanation for precognitive dreams but it shows where physics might be adjusted to accommodate the phenomenon,” he explains.Although Bierman’s explanation is still based on guesses and has not accepted by mainstream science, Watt does think it is worth considering. For now, believing that it’s possible to predict future with dreams remains an act of faith. Yet, it’s possible that one day we’ll wake up to a true understanding of this fascinating phenomenon. 63．According to French, what makes it difficult to test precognitive dreams?A．Unavailability of people’s dreams.B．That coincidences happen a lot in reality.C．That criteria for dream reliability are not trustworthy.D．People’s inability to tell when dreamt events will happen.64．Believers in precognitive dreams may question the truth of ______.A．the assumption of causality B．the time symmetryC．memories of ordinary people D．modern scientific tests65．We can infer from the passage that ______.A．Lincoln was warned of the killing by his friendB．Watt carried out several experiments on causalityC．researches on electromagnetism are based on the time symmetryD．time’s moving in two directions may justify precognitive dreams66．Which might be the best title of the passage?A．Should Dreams Be Assessed?B．Can Dreams Predict the Future?C．How Can Physics Be Changed to Explain Dreams?D．Why Should Scientists Study Precognitive Dreams?Section CDirections: Read the following passage. Fill in each blank with a proper sentence given in the box. Each sentence can only be used once. Note that there are two sentences more than you need.A．Labeling poses even more of a problem when it comes to kids.B．It can be helpful for those not quite able to understand why they feel the way they do.C．There seems to be a desire to see negative emotions as something requiring intervention or diagnosis. D．Labeling leads to children’s overcoming their addiction to what is posted online.E．Someone has had only a certain experience and judges all behavior with that experience.F．The basic function of a diagnosis is to give you a name for those behaviors once felt unusual.Addiction to LabelingMaybe you’ve noticed it in the comments section of popular social media posts about anxiety. depression or things alike, with a number of people claiming to pick these labels for themselves.These days, labeling is everywhere. （67）______ However, the negative part is that it’s easy for someone to identify with the characteristics without truly recognizing the context in which these characteristics would require diagnosis, according to Charlotte Armitage, a registered integrative psychotherapist and psychologist.If you have done your research and genuinely feel that you have some form of mental health concern, then finally having a name for your behaviors can be great. But the risk is that many people will seek labels and intervention for any behavior, pattern or emotion that is outside of the permanent happy group that society has set as the norm. “（68）______ Then the saying ‘a little bit of knowledge is dangerous’ springs to my mind,” Armitage adds.（69）______“Children are still developing and evolving, and many childhood behavioral features may seem like those of a disorder when there’re other potential explanations for that behavior,” Armitage notes. Ideally, a diagnosis for a child should be carried out by a qualified mental health professional. So it is with an adult.Nevertheless, the most important thing to bear in mind is that diagnosis doesn’t mean to indicate that you are broken or less capable.（70）______ And if you go deeper, it can alert you to the fact that you are not alone, and that many people experience life in the same way as you do.Ⅳ．Summary WritingDirections: Read the following passage. Summarize the main idea and the main point（s）of the passage in no more than 60 words. Use your own words as far as possible.71．Why Willing to Wait?First it was the fried chicken. Then a variety of fancy milkshakes. No matter what time it is or how bad the streets smell, there are plenty of people waiting in line for hours to get their hands on the food that everyone’s talking about. If you are not the type of person crazy for trendy foods, you probably wonder why someone would like to wait in a long line just to get a taste of a popular cream tea. There is a bit of psychology behind the craze of waiting before getting one’s chopsticks on a trendy food.People are born curiosity hunters, especially for fresh ideas, according to some experts. At the sight of a long waiting line, they just can’t help having a try. And when the trendy foods are novel in looks and favors, even innovative in their sales environment, the desire for them is upgraded. All those stimulate people to investigate more—to deal with their curiosity.In addition, having access to something that is sought out but hard to possess equips people with a feeling that improves their self-definitions. When someone is envied due to something he gained with efforts, his self-worth gets enhanced. Although it is yet to be determined whether the number of likes he receives on the photos of foods he’s posted online is connected with the level of envy from on-lookers, that feeling automatically becomes stronger.Even more, “mob psychology” comes into play: when many people are doing something—waiting in line for the sought-after milkshakes, for instance —others are eager to be part of the group and share such a type of social familiarity, kind of like the natural pursuit of a sense of belonging. Tasting the same wait-worthy food has something in common.Ⅴ．TranslationDirections: Translate the following sentences into English, using the words given in the brackets.72．大多数中国人喜欢在生日的时候吃碗面。
中国痴呆与认知障碍诊治指南写作组 英文
中国痴呆与认知障碍诊治指南写作组英文全文共10篇示例，供读者参考篇1Hey guys, do you know what dementia and cognitive impairment are? Today, I'm going to talk to you about it and how we can help those who have these conditions.Dementia is a condition that affects our brain and how we think, remember, and make decisions. It can make it hard for us to do everyday things like cooking, getting dressed, or even talking to our friends. Cognitive impairment is when our brain doesn't work as well as it should, and it can make it hard for us to learn new things or remember things we already know.But don't worry, there are ways to help people with dementia and cognitive impairment. One way is to encourage them to exercise their brains by doing puzzles, reading books, or even just talking to them and asking them questions. Another way is to make sure they eat well and get enough sleep, because a healthy body can help keep our brains healthy too.It's also important to be patient and understanding with people who have dementia or cognitive impairment. They mayforget things or get confused, but that's okay. Just be there for them and try to help them in any way you can.So let's all work together to support those who have dementia and cognitive impairment. We can make a difference in their lives and show them that they are not alone. Thanks for listening, and remember to be kind and compassionate to everyone you meet.篇2Title: A Guide to Understanding and Treating Dementia and Cognitive Impairment in ChinaHey guys, have you ever heard of something called dementia and cognitive impairment? It's a big word, but basically it means having trouble with your memory and thinking skills. It's something that can happen to older people, but it can also affect younger people too.In China, there are a lot of people who suffer from dementia and cognitive impairment. That's why it's important for us to learn about it and understand how we can help. There are many ways to diagnose and treat these conditions, so let's dive into some helpful tips and information!First of all, it's important to know the signs of dementia and cognitive impairment. Some common symptoms include forgetfulness, confusion, difficulty with everyday tasks, and changes in mood or behavior. If you or someone you know is experiencing these symptoms, it's important to see a doctor for a proper diagnosis.Once a diagnosis is made, there are different treatment options available. These can include medication, therapy, and lifestyle changes. It's important to work closely with healthcare professionals to find the best treatment plan for each individual.In addition to treatment, there are also ways to support people with dementia and cognitive impairment in their daily lives. This can include creating a safe and supportive environment, providing regular mental and physical stimulation, and maintaining a healthy diet and exercise routine.It's also important for families and caregivers to educate themselves about dementia and cognitive impairment. By understanding the condition and how to best support their loved ones, they can provide better care and improve quality of life.In conclusion, dementia and cognitive impairment are serious conditions that can impact people of all ages in China. By educating ourselves, seeking early diagnosis, and exploringtreatment options, we can help improve the lives of those affected by these conditions. Let's work together to create a more supportive and understanding community for all!篇3Hello everyone,Today, I want to talk to you about dementia and cognitive impairment. These are big words, but they are important to understand because they affect a lot of people, especially older people.Dementia is a condition where people have trouble remembering things, thinking clearly, and communicating. It can be really scary for someone to forget things or not be able to do things they used to do easily. Cognitive impairment is when people have trouble with their memory, attention, language, and reasoning skills. It's like their brain isn't working as well as it used to.But don't worry, there are ways to help people with dementia and cognitive impairment. They can go to the doctor to get a diagnosis and then they can get treatment to help them with their symptoms. There are also things that we can do tohelp them feel better, like spending time with them, talking to them, and helping them with everyday tasks.It's important to be patient and understanding with people who have dementia or cognitive impairment. They might get frustrated or confused, but we can help them by being kind and supportive. Let's all work together to make sure that everyone gets the help and support they need.Remember, we can all make a difference by being caring and understanding towards those who are facing these challenges. Let's show compassion and empathy to those who need it most.Thank you for listening and let's all do our part to support those with dementia and cognitive impairment. Together, we can make a positive impact on their lives.Take care and stay safe, everyone!Sincerely,[Your Name]篇4Title: Guide to Diagnosis and Treatment of Dementia and Cognitive Impairment in ChinaHey everyone! Today, let's talk about something super important - dementia and cognitive impairment. These are conditions that can affect our brains and make it hard for us to remember things or think clearly. But don't worry, we've got a guide to help you understand more about them and how to deal with them!First off, what is dementia? Dementia is a term used to describe a group of symptoms that affect our memory, thinking, and social abilities. It can be caused by different things like Alzheimer's disease or stroke. Cognitive impairment is similar, but it's a milder form of memory loss or trouble with thinking.So, how do we know if someone has dementia or cognitive impairment? Well, some signs to look out for include forgetting things often, having trouble with words or numbers, or getting lost in familiar places. If you notice any of these things in yourself or a loved one, it's important to see a doctor for a proper diagnosis.Once diagnosed, there are different treatments and therapies available to help manage dementia and cognitive impairment. These can include medications, cognitive therapy, and lifestyle changes like eating a healthy diet and staying active.In China, there are also resources available to help support those with dementia and cognitive impairment. There are specialized clinics and programs that provide care and assistance, as well as organizations that offer education and advocacy for those affected by these conditions.Remember, it's important to seek help and support if you or someone you know is experiencing symptoms of dementia or cognitive impairment. By working together and staying informed, we can better understand these conditions and help those affected live their best lives possible.That's all for today, folks! Stay sharp and take care of your brains!篇5Hi guys! Today I want to talk about China's dementia and cognitive impairment diagnosis and treatment guidelines writing group.So, first of all, let's talk about what dementia and cognitive impairment are. Dementia is when you have trouble remembering things, thinking clearly, or making decisions. It's like your brain is all mixed up and you can't do the things you used to do. Cognitive impairment is when your brain doesn'twork as well as it should. You might have trouble thinking, remembering, or learning new things.The China dementia and cognitive impairment diagnosis and treatment guidelines writing group is a group of smart people who study how to help people with dementia and cognitive impairment. They write down all the things that doctors should do to help people with these problems.One important thing they do is to make sure doctors can figure out if someone has dementia or cognitive impairment. They do tests and ask questions to see how well your brain is working. Then, they can give you the right medicine or therapy to help you feel better.It's really important to take care of our brains, guys! So if you or someone you know is having trouble remembering things or thinking clearly, make sure to go see a doctor. They can help you get better and feel like yourself again.Remember, it's okay to ask for help when you need it. The China dementia and cognitive impairment diagnosis and treatment guidelines writing group is here to help you!篇6Hello everyone! Today I'm going to talk about dementia and cognitive impairment in China. Do you know what that means? It's when people have trouble remembering things or thinking clearly. It can be really hard for them and their families.But don't worry, there are ways to help people with dementia and cognitive impairment. Doctors can give them medicine or therapy to make them feel better. They can also do things like puzzles or games to exercise their brains.It's important for us to be kind and patient with people who have dementia. They might get confused or frustrated, but we should always try to understand and help them as best as we can.If you know someone who has dementia or cognitive impairment, make sure to show them love and support. Spend time with them, listen to them, and try to make them feel happy.Let's all work together to make life better for people with dementia and cognitive impairment. We can make a difference by being compassionate and caring towards them. Thank you for listening!篇7Hello everyone! Today I want to talk to you about Chinese guidelines for the diagnosis and treatment of dementia and cognitive impairment. It's a big topic, but I'll try to break it down for you in simple terms.First of all, what is dementia? Dementia is a syndrome that affects memory, thinking, behavior and the ability to perform everyday activities. It is not a normal part of aging, and can be caused by various diseases or conditions.In China, the diagnosis of dementia is based on a comprehensive assessment that includes medical history, physical examination, cognitive tests, blood tests and brain imaging. Treatment usually involves a combination of medication, therapy and lifestyle changes.There are also guidelines for the management of specific types of dementia, such as Alzheimer's disease and vascular dementia. These guidelines provide recommendations on medication, therapy, and support for patients and their families.It's important to remember that early diagnosis and treatment of dementia can help improve quality of life and slow down progression of the disease. So if you or a loved one are experiencing memory problems or other symptoms of dementia, don't hesitate to seek help from a healthcare professional.That's all for today! Remember, knowledge is power, so stay informed and take care of your brain. Thanks for listening!篇8Hi everyone, today I'm going to talk about the Chinese Dementia and Cognitive Impairment Diagnosis and Treatment Guidelines Writing Group. It's a big word, I know, but it's important to understand how to take care of our brains!First of all, what is dementia and cognitive impairment? Well, it's when our brains start to have trouble with things like memory, thinking, and reasoning. It's like when you forget where you put your toys or what you had for breakfast. It can be really scary for people who have it, so it's important to know how to help them.The guidelines from the writing group give doctors and nurses information on how to diagnose and treat dementia and cognitive impairment. They can do things like memory tests and brain scans to see what's going on in the brain. They can also give medicines and therapy to help improve symptoms.It's also really important for us to take care of our brains every day. Things like eating healthy foods, exercising, and staying social can help keep our brains healthy. And if you noticesomeone having trouble with their memory or thinking, be kind and patient with them. They might need a little extra help.So let's all work together to learn more about dementia and cognitive impairment, and how we can help people who have it. Our brains are super important, so let's take care of them!篇9Hello everyone! Today, I'm going to talk about something super important - Chinese Dementia and Cognitive Impairment Diagnosis and Treatment Guidelines. Yeah, that's a mouthful, but don't worry, I'm here to break it down for you!So, what exactly is dementia and cognitive impairment? Well, it's basically when your brain doesn't work as well as it used to. It can make it hard to remember things, think clearly, or even do everyday tasks. But don't worry, there are ways to help!First off, it's important to see a doctor if you or someone you know is having trouble with their memory or thinking. They can do some tests to figure out what's going on and come up with a plan to help.One way to help with dementia and cognitive impairment is through lifestyle changes. Eating healthy, exercising, and stayingsocial can all help keep your brain in tip-top shape. Plus, it's important to keep your brain active by doing puzzles, reading, or learning new things.There are also medications that can help with symptoms of dementia and cognitive impairment. These can help improve memory, thinking, and even mood. Just make sure to talk to your doctor about any medications you're taking.And finally, it's important to have a good support system. Whether it's friends, family, or a support group, having people who care about you can make a big difference.So, remember, if you're worried about your memory or thinking, don't be afraid to talk to a doctor. There are ways to help improve your brain function and make life easier. Stay healthy, keep learning, and don't forget to take care of your brain!篇10Hello everyone! Today I want to tell you about the Chinese dementia and cognitive impairment diagnosis and treatment guidelines writing group. It's a group of smart people who are working hard to help patients with dementia and cognitive impairment in China.First of all, let's talk about what dementia and cognitive impairment are. Dementia is a condition that affects a person's memory, thinking, and behavior. It can make it difficult for someone to do everyday tasks and even recognize their loved ones. Cognitive impairment is when a person has trouble with their memory, attention, or problem-solving skills.The writing group is making guidelines to help doctors in China diagnose and treat people with dementia and cognitive impairment. They are working to improve the quality of care for these patients and make sure they get the help they need.The guidelines will provide doctors with important information on how to diagnose dementia and cognitive impairment. They will also give recommendations on the best ways to treat these conditions, such as medication, therapy, and lifestyle changes.Overall, the Chinese dementia and cognitive impairment diagnosis and treatment guidelines writing group is doing important work to help people in China who are struggling with these conditions. Let's give them a big round of applause for all their hard work!Remember, if you or someone you know is experiencing memory problems or other symptoms of dementia, it's important to see a doctor for help. Don't wait, take action now!。
阿尔茨海默病的生信文章
阿尔茨海默病的生信文章阿尔茨海默病是一种以失忆、认知和行为障碍为主要表现的神经系统退行性疾病。
生信技术在阿尔茨海默病的研究中发挥着重要作用。
以下是一些相关的生信文章：1. “Identifying the Genetic Risk Factors forAlzheimer's Disease through Genome-Wide Association Studies”. 该文章介绍了基因组关联研究（GWAS）在阿尔茨海默病研究中的应用，了解了遗传因素对该疾病的风险的贡献。
2. “RNA sequencing-based transcriptome profiling of Alzheimer's disease patients”. 该文章介绍了RNA测序技术在阿尔茨海默病研究中的应用，通过对病人和正常人脑组织中RNA表达谱的比较和分析，揭示了该疾病的复杂机制。
3. “Proteomics in Alzheimer's disease: insights into potential biomarkers and therapeutic targets”. 该文章介绍了蛋白质组学在阿尔茨海默病研究中的应用，以期寻找潜在的生物标志物和治疗靶点。
4. “Machine learning approaches for predictingAlzheimer's disease: a review”. 该文章介绍了机器学习技术在阿尔茨海默病的预测方面的应用，探讨该技术在该疾病早期诊断和治疗方面的潜力。
5. “Single-cell analysis of Alzheimer's disease”. 该文章利用单细胞分析技术探讨了阿尔茨海默病中不同细胞类型和亚型的转录组和表观遗传表达差异，以期深入了解该疾病的发病机制和治疗方式。
总之，生信技术在阿尔茨海默病研究中的应用将为我们更好地理解该疾病的发病机制，开发更有效的治疗方法提供有力保障。
网络流行语“我的内心几乎是崩溃的”的模因论解读
校园英语 / 语言文化网络流行语“我的内心几乎是崩溃的”的模因论解读西北师范大学外国语学院/范攀攀【摘要】2015年“我的内心几乎是崩溃的”蹿红网络，成为大众追捧的网络流行语。
模因论对于此类语言现象具有很强的解释力。
本文从“我的内心几乎是崩溃的”的释义及产生入手，从模因论的角度出来研究其复制和传播的过程，探讨了其成为强势模因的语言内在和外在的原因，以期为运用模因论分析流行语提供借鉴和参考。
【关键词】我的内心几乎是崩溃的 模因 强势模因【Abstract】In 2015, “我的内心几乎是崩溃的”(My heart is almost broken)became extremely popular among netizens. Memetics theory has a strong explanatory power for this kind of language phenomenon. This paper begins by explaining the meaning and tracking the source of the popular expression. Based on an analysis of how the expression replicates and spreads, the paper further explores language reasons -- both internal and external -- for its becoming a powerful meme in light of memetics. This paper attempts to provide people with some potential insights to effectively analyze other popular expressions from the perspective of memetics.【Key words】My heart is almost broken; meme; powerful meme一、引言近年来，随着互联网的快速发展，网络语言逐渐走进我们的生活。
一种新的部分神经进化网络的股票预测(英文)
一种新的部分神经进化网络的股票预测(英文)一种新的部分神经进化网络的股票预测自从股票市场的出现以来，人们一直在寻求能够提前预测股票走势的方法。
许多投资者和研究人员尝试使用各种技术分析工具和模型来预测股票未来的走势，但是股票市场的复杂性和难以预测性使得这变得困难重重。
因此，寻找一种能够准确预测股票走势的方法一直是金融界的热点问题。
近年来，人工智能技术在金融领域的应用日益增多。
其中，神经网络是一种被广泛使用的工具，它可以自动学习和识别模式，并根据所学的模式进行预测。
然而，传统神经网络在预测股票市场方面存在诸多问题，例如过拟合和难以处理大量数据等。
为了克服这些问题，本文提出了一种新的部分神经进化网络（Partial Neural Evolving Network, PNEN）模型来预测股票走势。
PNEN模型将神经网络和进化算法相结合，通过优化和训练来实现更准确的预测结果。
PNEN模型的核心思想是将神经网络的隐藏层拆分为多个小模块，每个小模块只负责处理一部分输入数据。
通过这种方式，模型可以更好地适应不同的市场情况和模式。
同时，采用进化算法来优化模型的参数，可以进一步提高模型的预测性能。
具体而言，PNEN模型包括以下几个步骤：1. 数据准备：从股票市场获取历史交易数据，并对数据进行预处理和归一化处理，以便更好地输入到模型中。
2. 构建模型结构：将神经网络的隐藏层拆分为多个小模块，通过进化算法来确定每个小模块的结构和参数。
进化算法通过优化模型的准确性和稳定性，以获得更好的预测结果。
3. 训练模型：使用历史数据集对模型进行训练，并通过反向传播算法来更新模型的权重和偏置。
同时，通过与进化算法的交互，不断调整模型结构和参数。
4. 预测结果：使用训练好的模型对未来的股票走势进行预测。
通过模型对市场的分析和判断，可以为投资者提供决策参考。
为了验证PNEN模型的效果，我们在实际的股票市场数据上进行了实验。
结果表明，与传统神经网络模型相比，PNEN 模型在预测股票走势方面具有更好的准确性和稳定性。
关于一知半解的开头作文
关于一知半解的开头作文英文回答：Half-knowledge, often referred to as "pseudo-knowledge" or "superficial understanding," poses significant risks and drawbacks. It can lead to:1. Misguided Decisions: Individuals with half-knowledge may make hasty and uninformed decisions, often based on incomplete or inaccurate information. This can have detrimental consequences for themselves and others.2. Intellectual Stagnation: Half-knowledge can create a false sense of accomplishment, leading to complacency and hindering the pursuit of genuine understanding. Individuals may believe they know enough, preventing them from exploring deeper insights or seeking further knowledge.3. Inability to Critically Evaluate: Those with half-knowledge often lack the depth of understanding necessaryto critically evaluate information. They may blindly accept claims or ideas without considering their validity, making them susceptible to misinformation and manipulation.4. Confidence Trap: Half-knowledge can create a false sense of confidence, leading individuals to overestimate their capabilities. This overconfidence can hinder their ability to recognize their own limitations and seek expert advice when necessary.5. Communication Barriers: Half-knowledge can interfere with effective communication. Individuals with different levels of understanding may struggle to engage in meaningful discussions or convey complex ideas accurately.6. Misrepresentation of Knowledge: Those with half-knowledge may inadvertently misrepresent their level of expertise or spread inaccurate information, unknowingly contributing to the dissemination of misinformation.To overcome the risks associated with half-knowledge, it is essential to:1. Seek Comprehensive Understanding: Pursue a thorough understanding of subjects and avoid relying solely on superficial knowledge. This requires active learning,critical thinking, and a willingness to delve into complexities.2. Recognize Limitations: Acknowledge the boundaries of one's knowledge and seek expert guidance when necessary. Understand that it is okay to admit when one does not knowor fully understand something.3. Develop Critical Thinking Skills: Cultivate theability to analyze information objectively, identify biases, and evaluate the credibility of sources. This involves questioning assumptions, examining evidence, and engagingin logical reasoning.4. Embrace Curiosity and Lifelong Learning: Maintain a mindset of intellectual curiosity and a commitment to continuous learning. Stay open to new ideas and experiences, recognizing that knowledge evolves and deepens over time.5. Promote Responsible Knowledge Sharing: Be mindful of the potential risks of disseminating half-knowledge. Share information responsibly, ensuring its accuracy and completeness. Encourage others to pursue deeper understanding and critical evaluation.中文回答：一知半解，常称为“似是而非的知识”或“肤浅的理解”，具有显着的风险和弊端。
基于深度学习和蛋白质语言模型的抗菌肽预测模型研究
㊀㊀㊀㊀㊀㊀㊀㊀㊀2024年4月第39卷第2期JOURNAL OF LIGHT INDUSTRY㊀Vol.39No.2Apr.2024㊀收稿日期:2023-10-19;修回日期:2024-01-25;出版日期:2024-04-15基金项目:国家自然科学基金青年科学基金项目(32101976,61906175);河南省科技攻关项目(232102210020,20210221014);河南省高等学校重点科研项目(22A520013,23B520004)作者简介:王晓(1982 ),男,河南省卫辉市人,郑州轻工业大学副教授,博士,主要研究方向为人工智能㊁生物信息学和食品信息学㊂E-mail :wangxiao@.cn王晓,吴洲,王宏伟,等.基于深度学习和蛋白质语言模型的抗菌肽预测模型研究[J].轻工学报,2024,39(2):12-18.WANG X,WU Z,WANG H W,et al.Research on antimicrobial peptide prediction model based on deep learning and protein language model[J].Journal of Light Industry,2024,39(2):12-18.DOI:10.12187/2024.02.002基于深度学习和蛋白质语言模型的抗菌肽预测模型研究王晓1,2,吴洲1,王宏伟3,王榕4,陈浩然11.郑州轻工业大学计算机科学与技术学院,河南郑州450001;2.河南省食品安全数据智能重点实验室,河南郑州450001;3.郑州轻工业大学食品与生物工程学院,河南郑州450001;4.郑州轻工业大学电子信息学院,河南郑州450001摘要:针对目前已有抗菌肽(Antimicrobial Peptides ,AMPs )预测模型的准确度(ACC )仍有待提高的问题,提出一种新的基于深度学习和蛋白质语言模型的抗菌肽预测模型DeepGlap ,该模型分别采用两个蛋白质语言模型对抗菌肽序列进行特征提取,将提取的特征向量融合后输入由多层双向长短记忆网络(mBi-LSTM )㊁一维卷积神经网络(1D-CNN )和注意力机制组成的深度学习网络中,并进行性能评估与优化㊂结果表明:该模型的ACC ㊁皮尔逊相关系数(MCC )和曲线下的面积(AUC )分别为0.739㊁0.489和0.81,优于已有抗菌肽预测模型的预测效果㊂关键词:抗菌肽;预测模型;食源性病原体;蛋白质语言模型;深度学习网络中图分类号:TS201.2;TP399㊀㊀文献标识码:A㊀㊀文章编号:2096-1553(2024)02-0012-070 引言食源性病原体是导致食品安全问题的重要来源之一,研究[1]表明,在畜禽养殖过程中使用抗生素会产生具有耐药性的人畜共患食源性病原体㊂抗菌肽(Antimicrobial Peptides,AMPs)作为抗生素的一种潜在替代品,是一类广泛存在于动植物体内的具有抗菌活性的小分子蛋白质或多肽,通过破坏细胞膜㊁干扰蛋白质合成㊁促使微生物发生自我毁灭等方式杀灭细菌㊁真菌㊁病毒等微生物或抑制这些微生物生长繁殖[2]㊂抗菌肽的鉴别有助于减轻抗生素依赖,降低因畜禽养殖过度使用抗生素而带来的耐药性病原体风险,对于研发新型抗菌剂㊁有效保障食品安全和维护公共健康具有重要意义㊂然而,通过湿实验来鉴别抗菌肽操作复杂,且耗费大量的时间与资源[3],因此有必要借助现代计算科学手段开发更高效的抗菌肽预测模型㊂目前,预测抗菌肽的计算方法主要有:1)传统的机器学习方法㊂其中,支持向量机㊁随机森林㊁XGBOOST 及集成学习都得到了广泛应用,如㊃21㊃㊀王晓,等:基于深度学习和蛋白质语言模型的抗菌肽预测模型研究ClassAMP使用随机森林和支持向量机实现了抗菌肽的预测[4];iAMPpred引入蛋白质序列的结构特征和理化性质,并与传统手工特征合并后输入支持向量机以进行抗菌肽的分类预测,可进一步提高抗菌肽的预测精度[5];IAMPE利用氨基酸的CNMR光谱提取特征向量,并与蛋白质的传统手工特征向量融合后输入支持向量机和随机森林来实现抗菌肽的分类预测,该方法获得了较好的预测精度[6];AMPfun 则通过提取序列的n-gram特征㊁AAC特征㊁理化性质等特征向量,并输入支持向量机中以实现抗菌肽的分类预测,进一步提高了抗菌肽的预测精度[7]㊂2)新型的机器学习方法㊂如iAMPCN通过不同的卷积神经网络(CNN)提取4种不同编码方法的特征向量,在实现抗菌肽分类预测的基础上,首次实现了抗菌肽的功能预测[8];sAMPpred-GAT使用图注意力机制,将结构特征引入深度学习网络中,进一步提高了抗菌肽的预测精度[9]㊂大语言模型是能够基于原有文本来预测下一个字符或词汇的一类神经网络,而蛋白质语言模型是大语言模型在生化领域的迁移应用,在输入蛋白质序列并学习序列中隐含的生化性质㊁结构信息等内在规律后,生成蛋白质的特征向量,这些特征向量可用于预测多种蛋白质属性㊂研究[10]表明,多种下游任务使用蛋白质语言模型进行预测,均取得了较好的效果㊂基于此,本文拟设计一种新的基于深度学习和蛋白质语言模型的抗菌肽预测模型DeepGlap,将两个蛋白质语言模型所生成的特征向量融合后,输入由多层双向长短记忆网络(mBi-LSTM)㊁一维卷积神经网络(1D-CNN)和注意力机制所组成的深度学习网络中,并对该模型的准确性进行验证,以期获得更好的抗菌肽预测效果㊂1㊀数据集与数据预处理迄今为止,已建立了大量的抗菌肽数据库,其中APD(Antimicrobial Peptide Database)作为早期建立的抗菌肽数据库之一,包含不同生物界(细菌㊁真菌㊁动物等)的抗菌肽数据,以及这些抗菌肽的分类㊁结构㊁活性等信息[11];LAMP(Linking Antimicro-bial Peptide)提供了各种生物体内外来源的抗菌肽序列信息,以及相关的文献引用和其他注释数据[12];CAMP(Collection of Anti-Microbial Peptides)汇集了来自不同物种的抗菌肽信息,包含各种结构分类信息[13];DBAASP(Database of Antimicrobial Activity and Structure of Peptides)是一个用于存储和提供抗菌肽信息的数据库,包括抗菌肽的序列㊁结构㊁抗菌活性㊁相关文献引用等信息[14];DRAMP (Data Repository of Antimicrobial Peptides)是一个综合的抗菌肽数据库,包含抗菌肽的结构数据㊁注释条目等[15];SATPdb(Database of Structurally Annotated Therapeutic Peptides)可提供大量的抗菌肽结构数据,其主要部分是通过计算工具预测获取的[16]㊂这些数据库的建立为研究人员提供了便利,有助于推动抗菌肽领域的深入研究㊂为了消除不同数据库数据分布不同所带来的影响,本文将一个综合基准评测数据集用于模型的训练和验证,该数据集包含了来自APD㊁LAMP㊁CAMP㊁DBAASP㊁DRAMP和SATPdb这6个数据库的抗菌肽和非抗菌肽数据㊂由于新数据库可能参考了较早数据库的数据,不同数据库之间的数据可能有重叠,因此本文采用CD-HIT工具[17]消除冗余数据,即当来自不同数据库的肽序列之间的相似度高于90%时,将其中一条数据删除㊂根据Y.Zhang 等[18]和K.Yan等[9]的研究,肽链过长可能会产生更复杂的结构,导致蛋白质难以合成,而肽链过短又可能无法包含足够的功能性位点或结构域,且在环境中易发生降解[19],故本文选择长度为10~100个氨基酸的序列,并删除包含非标准氨基酸(B,J,O, U)的序列,最终获得包含4550条抗菌肽和4550条非抗菌肽的基准评测数据集㊂本文采用CD-HIT工具删除独立测试集XUA-MP(包含1536条抗菌肽和1536条非抗菌肽)[19]与基准评测数据集中相似度高于90%的肽序列,确保数据的独立性,从而更客观地评价预测模型的泛化能力㊂2㊀特征提取传统手工特征提取方法通常依赖专家知识提取蛋白质序列的生化统计信息,无法全面准确地反映㊃31㊃㊀2024年4月第39卷第2期㊀蛋白质序列的复杂模式特征;而蛋白质语言模型通过无监督深度学习技术,学习数亿条蛋白质序列的复杂依赖关系和结构信息,可以全面准确地提取蛋白质序列的特征㊂因此,本文使用蛋白质语言模型Unirep和Protbert对抗菌肽序列分别进行特征提取,并将提取的特征向量进行融合,从而获得更全面的抗菌肽特征向量㊂2.1㊀Unirep蛋白质语言模型Unirep蛋白质语言模型采用长短时记忆网络(Long Short Memory Network,LSTM)作为基础架构,通过预测序列中下一个氨基酸值是否与真实氨基酸值相同,不断优化LSTM,最终使用多个LSTM的隐藏层单元平均值作为序列的特征向量㊂Unirep蛋白质语言模型使用多个图形处理器(Graphics Process-ing Unit,GPU),在UniRef50蛋白质数据库中约2400万条蛋白质序列上进行为期3周的训练[20],最终将不同长度的蛋白质序列映射为统一长度的1900维特征向量,通过这种方式可有效将具有较低统一性的蛋白质序列划分为结构相似性较高的类别㊂2.2㊀ProtBert蛋白质语言模型ProtBert蛋白质语言模型作为Transformer/ BERT架构的创新模型,使用来自BDF蛋白质数据库和UniRef50蛋白质数据库中超过2ˑ108条蛋白质序列进行广泛的训练,最终实现将不同长度的蛋白质序列映射为统一长度的1024维特征向量㊂该模型汲取了Transformer架构中的多头注意力机制,在较高水平全局特征的基础上能更加突出蛋白质序列的局部特征[21]㊂3㊀深度学习网络模型抗菌肽预测模型的预测流程为:首先,将抗菌肽序列分别输入蛋白质语言模型Unirep和ProtBert中,得到对应的1900维特征向量和1024维特征向量,并将这两个特征向量进行特征融合;随后,将融合的特征向量依次输入由mBi-LSTM㊁1D-CNN和注意力机制组成的深度学习网络中;最后,输入由Flatten操作㊁批归一化(Batch-Normalize)和Sigmoid 激活函数组成的输出层中,完成抗菌肽的分类预测㊂本文所提出的模型采用ReduceLROnPlateau动态学习率算法,该算法是深度学习框架PyTorch中的学习率调度器之一,可在训练过程中根据验证集上的性能指标动态调整学习率,如在验证集性能停止提升时降低学习率,以帮助模型更好地收敛㊂具体而言,当测试集的准确度(ACC)连续两个epoch 保持不变时,ReduceLROnPlateau动态学习率算法会将学习率调整为原学习率的0.7倍,直到达到最大epoch值为止㊂本文在模型训练过程中采用二分类交叉熵损失函数,损失(Loss)的计算方法如式①㊁式②所示㊂Loss=1NðN n=1l n①l n=y n㊃log x n+(1-y n)㊃log(1-x n)②式中,l n是单个样本的损失值,n是当前样本标号, N是样本数量,x n是模型输出,y n是真实标签㊂本文利用ADAM算法优化模型,批量大小为64,初始学习率为0.001,并采用十折交叉验证,每一折包含20个epoch㊂3.1㊀mBi-LSTM循环神经网络(Rerrent Neural Network,RNN)已广泛应用于处理时间序列和文本数据,但在处理较长序列数据时,存在无法提取序列的长期依赖的限制㊂为了解决这个问题,学者们将LSTM作为RNN的一种变体㊂LSTM引入细胞状态的概念,以保持前后时刻之间的信息流动,并通过遗忘门㊁输入门和输出门这3个门结构对输入信息和隐藏状态进行调控㊂然而,传统RNN和LSTM在信息传递方面都是单向的,无法获取未来状态的信息㊂蛋白质序列可类比为生物学上的一种语言:肽段相当于句子,而氨基酸残基则类似于单词㊂因此,氨基酸残基之间的上下文关系对预测结果至关重要㊂为了弥补LSTM 的局限性,本文采用mBi-LSTM从相反的方向学习传统特征向量的输入,在不同方向上捕捉序列中的关键模式和信息,从而更全面地表征蛋白质序列,更好地学习和捕获更复杂序列数据中的依赖关系㊂3.2㊀1D-CNN1D-CNN作为深度学习中提取一维数据的网络㊃41㊃㊀王晓,等:基于深度学习和蛋白质语言模型的抗菌肽预测模型研究架构,可用于序列数据的特征提取㊂1D-CNN能接收来自mBi-LSTM层的输出向量,捕捉序列中的局部特征,并通过堆叠多个卷积层,学习更高级㊁更抽象的特征向量㊂3.3㊀注意力机制注意力机制最初是为了解决机器翻译任务而提出的,在自然语言处理领域有着广泛的应用㊂随着研究的不断深入,注意力机制在计算机视觉领域和图像处理领域均取得了突出的成果,其善于从输入数据中挖掘重要信息,并有效地集成多视图特征[22]㊂此外,注意力机制常与RNN相结合用于生物信息学,且在生物序列分析问题中取得了较高的性能得分[23]㊂3.4㊀输出层输出层的作用是将神经网络处理后的特征向量进行变形㊁加工,最终输出Flatten操作㊁批归一化㊁Sigmoid激活函数等分类结果㊂其中,Flatten是神经网络中的一种操作,通常用于从多维输入数据中拉平数据维度,以便将其作为全连接层的输入㊂在本文中,由1D-CNN输出的向量维度为3维,通过Flat-ten可将其变成2维,以便于输入全连接层进行后续分类处理㊂批归一化的主要思想是在每个训练批次的数据上进行归一化,以确保输入数据的分布稳定性㊂具体操作为:计算每个批次数据的平均值和标准差,然后对数据进行归一化,使其平均值为0,标准差为1㊂计算方法如式③所示㊂x i+1=x i-μBσBα+β③式中,x i+1是批归一化后的数据,μB是数据样本的平均值,σB是数据样本的标准差,α和β分别是对数据样本进行的拉伸和偏移,i是x的时间步㊂对特征向量进行所有处理后,通过线性层对其进行分类,验证是否为抗菌肽㊂线性层有较高的分辨能力,如式④所示㊂x t=W t l x(t-1)+b t④式中,x t和x(t-1)分别是输出向量和输入向量,W t l是权重矩阵,b t是线性层的偏差,t㊁l是时间步序号㊂除第一层和最后一层外,将批归一化用于每个线性层的输出;将Sigmoid激活函数用于最后一层的输出㊂Sigmoid激活函数可将输出转化为0~1,并以此作为概率,认为ȡ0.5的输出为抗菌肽,<0.5的输出为非抗菌肽㊂Sigmoid激活函数如式⑤所示㊂S(x)=11+e-x⑤式中,x为最后一层的输出㊂4㊀模型性能评估与优化4.1㊀性能评估通过对模型性能进行评估,可在众多参数的组合中选出最适合该预测模型的参数组合,从而有效预测抗菌肽㊂在本文中利用5个指标来评估模型性能,其计算公式如式⑥所示㊂ACC=TP+TNTP+TN+FN+FPMCC=TPˑTN-FNˑFP(TP+FN)(TP+FP)(TN+FP)(TN+FN)Sn=TPTP+TNSp=TNFP+TNìîíïïïïïïïïïïï⑥式中,TP㊁FP㊁TN和FN分别是真阳性㊁假阳性㊁真阴性和假阴性;当ACC=1时,说明所有的抗菌肽预测正确,当ACC=0时,说明所有的抗菌肽预测错误;MCC是皮尔逊相关系数,MCC越接近1,则表示模型综合性能越高,MCC越接近0,则表示模型性能越接近随机分类器,MCC越接近-1,则表示模型的综合性能越差;Sp是模型的特指度,Sn是模型的灵敏度,二者分别代表模型预测抗菌肽和非抗菌肽的能力㊂4.2㊀模型优化为了使融合向量能更好地反映抗菌肽特征,本文在使用基准评测数据集进行训练和验证时,尝试不同的特征向量融合方法,包括自注意力机制融合特征向量㊁交叉注意力机制融合特征向量和向量拼接,并采用十折交叉验证来保证实验结果的可靠性㊂不同特征向量融合方法的效果见表1㊂由表1可知,向量拼接法对应的ACC最高,模型预测效果最好㊂4.3㊀参数优化4.3.1㊀mBi-LSTM层参数优化㊀为了使mBi-LSTM㊃51㊃㊀2024年4月第39卷第2期㊀达到最优效果,训练过程中在基准数据集上进行mBi-LSTM层参数优化,并采用十折交叉验证来保证实验结果的可靠性㊂mBi-LSTM的层数分别设置为1~7层,不同mBi-LSTM层参数优化的效果见表2㊂由表2可知,当mBi-LSTM层数设置为2层时, ACC最高,模型预测效果最好㊂4.3.2㊀1D-CNN层参数优化㊀为了使1D-CNN层能更好地处理mBi-LSTM层的输出向量,训练过程中在基准评测数据集上进行1D-CNN层的参数优化,并使用十折交叉验证来保证实验结果的可靠性㊂1D-CNN层的输入通道数分别设置为2个㊁4个㊁6个㊁8个和10个,卷积核感受野大小分别设置为1600㊁1800㊁2000㊁2200和2400,不同通道数模型及不同卷积核感受野大小模型的效果分别见表3和表4㊂由表3和表4可知,输入通道为2个㊁卷积核感受野大小为2000的模型ACC最高,模型预测效果最好㊂表1㊀不同特征向量融合方法的效果Table1㊀Effect of different teature vectorfusion methods方法ACC MCC Sp Sn自注意力机制0.640.300.750.54交叉注意力机制0.500.00 1.000.00向量拼接0.740.480.630.84表2㊀不同mBi-LSTM层参数优化的效果Table2㊀Effect of optimizing parametersof different mBi-LSTM layers层数/层ACC MCC Sp Sn 10.700.410.830.56 20.740.490.840.64 30.680.380.610.75 40.700.410.670.74 50.710.440.850.56 60.710.430.740.67 70.720.440.790.63表3㊀不同通道数模型的效果Table3㊀Effect of models with differentchannel number通道数/个ACC MCC Sp Sn20.730.490.840.6440.690.430.900.4960.700.420.800.6080.700.410.730.67100.710.420.750.665㊀实验结果分析本文将所提出的模型DeepGlap与10种较先进的预测模型(包括amPEPpy㊁AMPfun㊁AMPEP㊁ADAM-HMM㊁AMPIR㊁AMPScannerV2㊁AMPGram㊁Deep-AMPEP30㊁CAMP-ANN㊁sAMPpred-GAT)在独立测试集XUAMP上进行比较,结果见表5㊂由表5可知,本文所提出的模型DeepGlap在ACC㊁MCC和Sn方面均有较好的性能,表明模型的准确率和泛化能力均达到了较高水平㊂而模型Deep-AMPEP30的Sp虽达到了1.000,但Sn非常小,说明该模型大幅度偏向于预测正样本,综合预测精度较差㊂为了进一步验证模型的优劣性,本文采用ROC (Receiver Operating Characteristic)曲线下的面积(Area Under the Curve,AUC)来评估模型的性能, AUC越大,表示模型的性能越好,误报率越低,反之模型性能越差,误报率越高㊂不同模型的AUC如图1所示㊂由图1可知,本文所提出的模型DeepGlap 的AUC最大(为0.81),表明该模型拥有更好的预测能力,且误报率较低㊂表4㊀不同卷积核感受野大小模型的效果Table4㊀Effect of models with different convolutionkernel receptive field sizes卷积核感受野大小ACC MCC Sp Sn 16000.660.390.690.7018000.660.400.940.3920000.700.490.840.6322000.710.430.790.6224000.700.420.840.56表5㊀不同模型在独立测试集XUAMP上的效果Table5㊀Effect of different models on theindependent test set XUAMP模型ACC MCC Sn SpamPEPpy0.6790.4310.4000.958AMPfun0.6740.4140.4060.943AMPEP0.6610.4290.3300.992 ADAM-HMM0.6840.3900.5210.847AMPIR0.5630.1560.2660.859 AMPScannerV20.5680.1370.5230.613 AMPGram0.5640.1310.4450.682 Deep-AMPEP300.5330.1830.065 1.000 CAMP-ANN0.5840.1820.3850.782 sAMPpred-GAT0.7150.4640.5300.900DeepGlap0.7390.4890.6350.843㊃61㊃㊀王晓,等:基于深度学习和蛋白质语言模型的抗菌肽预测模型研究图1㊀不同模型的AUCFig.1㊀The AUC of different models6㊀结论本文针对抗菌肽的精准鉴别问题,提出了一种新的基于深度学习和蛋白质语言模型的抗菌肽预测模型DeepGlap,该模型使用蛋白质语言模型Unirep 和ProtBert 提取抗菌肽的特征,再使用包含mBi-LSTM㊁1D-CNN 和注意力机制的深度学习网络进行堆叠,并对模型参数进行优化,得到如下结论:相较于已有抗菌肽预测模型,该抗菌肽预测模型的预测效果更佳,有望成为一个有前景的抗菌肽预测工具,以辅助新型抗菌剂的研发,降低耐药性病原体的感染风险㊂参考文献:[1]㊀WHITE D G ,ZHAO S H ,SIMJEE S ,et al.Antimicrobialresistance of foodborne pathogens [J ].Microbes and Infection ,2002,4(4):405-412.[2]㊀KUMAR P ,KIZHAKKEDATHU J N ,STRAUS S K.Anti-microbial peptides :Diversity ,mechanism of action and strategies to improve the activity and biocompatibility in vivo[J ].Biomolecules ,2018,8(1):4.[3]㊀ZHANG R N ,WU D ,GAO Y T.Progress on the designand optimization of antimicrobial peptides [J ].Journal of Biomedical Engineering ,2022,39(6):1247-1253.[4]㊀JOSEPH S ,KARNIK S ,NILAWE P ,et al.ClassAMP :Aprediction tool for classification of antimicrobial peptides[J ].IEEE-ACM Transactions on Computational Biology and Bioinformatics ,2012,9(5):1535-1538.[5]㊀SHARMA R ,SHRIVASTAVA S ,SINGH S K ,et al.Ani-AMPpred :Artificial intelligence guided discovery of novelantimicrobial peptides in animal kingdom [J ].Briefings inBioinformatics ,2021,22(6):bbab242.[6]㊀KAVOUSI K ,BAGHERI M ,BEHROUZI S ,et al.IAMPE :NMR-assisted computational prediction of antimicrobialpeptides [J ].Journal of Chemical Information and Model-ing ,2020,60(10):4691-4701.[7]㊀CHUNG C R ,KUO T R ,WU L C ,et al.Characterizationand identification of antimicrobial peptides with differentfunctional activities [J ].Briefings in Bioinformatics ,2020,21(3):1098-1114.[8]㊀XU J ,LI F Y ,LI C ,et al.iAMPCN :A deep-learningapproach for identifying antimicrobial peptides and theirfunctional activities [J ].Briefings in Bioinformatics ,2023,24(4):bbad240.[9]㊀YAN K ,LYU H W ,GUO Y C ,et al.sAMPpred-GAT :Prediction of antimicrobial peptide by graph attention net-work and predicted peptide structure [J ].Bioinformatics ,2023,39(1):btac715.[10]FERRUZ N ,HÖCKER B.Controllable protein design withlanguage models [J ].Nature Machine Intelligence ,2022,4(6):521-532.[11]WANG G S ,LI X ,WANG Z.APD3:The antimicrobialpeptide database as a tool for research and education [J ].Nucleic Acids Research ,2016,44:D1087-D1093.[12]ZHAO X W ,WU H Y ,LU H R ,et MP :A databaselinking antimicrobial peptides [J ].PLoS ONE ,2013,8(6):e66557.[13]THOMAS S ,KARNIK S ,BARAI R S ,et al.CAMP :Auseful resource for research on antimicrobial peptides [J ].Nucleic Acids Research ,2010,38:D774-D780.[14]GOGOLADZE G ,GRIGOLAVA M ,VISHNEPOLSKY B ,et al.DBAASP :Database of antimicrobial activity andstructure of peptides [J ].FEMS Microbiology Letters ,2014,357(1):63-68.[15]KANG X Y ,DONG F Y ,SHI C ,et al.DRAMP 2.0,anupdated data repository of antimicrobial peptides [J ].Sci-entific Data ,2019,6:148.[16]JHONG J H ,CHI Y H ,LI W C ,et al.dbAMP :An inte-grated resource for exploring antimicrobial peptides with functional activities and physicochemical properties on transcriptome and proteome data [J ].Nucleic Acids Research ,2019,47:D285-D297.[17]FU L M ,NIU B F ,ZHU Z W ,et al.CD-HIT :Accelerated for clustering the next-generation sequencing data [J ].Bioinformatics ,2012,28(23):3150-3152.[18]ZHANG Y ,LIN J Y ,ZHAO L M ,et al.A novel antibacte-rial peptide recognition algorithm based on BERT [J ].Briefings in Bioinformatics ,2021,22(6):bbab200.[19]XU J ,LI F Y ,LEIER A ,et prehensive assessmentof machine learning-based methods for predicting antimi-crobial peptides [J ].Briefings in Bioinformatics ,2021,22㊃71㊃㊀2024年4月第39卷第2期㊀(5):bbab083.[20]ALLEY E C ,KHIMULYA G ,BISWAS S ,et al.Unifiedrational protein engineering with sequence-based deeprepresentation learning [J ].Nature Methods ,2019,16(12):1315-1322.[21]BRANDES N ,OFER D ,PELEG Y ,et al.ProteinBERT :Auniversal deep-learning model of protein sequence and function [J ].Bioinformatics ,2022,38(8):2102-2110.[22]HOCHREITER S ,SCHMIDHUBER J.Long short-termmemory [J ].Neural Computation ,1997,9(8):1735-1780.[23]YU Z X ,HUANG F ,ZHAO X H ,et al.Predicting drug-disease associations through layer attention graph convolu-tional network [J ].Briefings in Bioinformatics ,2021,22(4):bbaa243.Research on antimicrobial peptide prediction model based on deep learningand protein language modelWANG Xiao 1,2,WU Zhou 1,WANG Hongwei 3,WANG Rong 4,CHEN Haoran 11.College of Computer Science and Technology ,Zhengzhou University of Light Industry ,Zhengzhou 450001,China ;2.Henan Key Laboratory of Data Intelligence for Food Safety ,Zhengzhou 450001,China ;3.College of Food and Bioengineering ,Zhengzhou University of Light Industry ,Zhengzhou 450001,China ;4.College of Electronic Information ,Zhengzhou University of Light Industry ,Zhengzhou 450001,ChinaAbstract :In response to the need for improving prediction accuracy (ACC )in existing models for AntimicrobialPeptides (AMPs),a novel AMP prediction model called DeepGlap was proposed.This model utilized two proteinlanguage models for feature extraction from AMP sequences,followed by fusion of feature vectors.These fusedvectors were then input into a deep learning network composed of multiple layers of bidirectional long short-term memory networks (mBi-LSTM ),one-dimensional convolutional neural networks (1D-CNN ),and attentionmechanisms.The model underwent performance evaluation and optimization.Results indicated that the model achieved ACC ,the Pearson correlation coefficient (MCC ),and the area urder the curve (AUC )values of 0.739,0.489,and 0.81,respectively,demonstrating superior predictive performance compared to existing AMPprediction models.Key words :antimicrobial peptide;prediction model;foodborne pathogen;protein language model;deep learningnetwork㊀[责任编辑:杨晓娟]本刊数字网络传播声明本刊已许可中国知网,万方数据资源系统,维普网,国家科技学术期刊开放平台,博看网,超星,中国科技论文在线,中教数据库,长江文库,CABI,CA,AJ.VINITI,EBSCO host ,Ulrichs,FSTA 等在其系列数据库产品中以数字化方式复制㊁汇编㊁发行㊁信息网络传播本刊全文㊂其相关著作权使用费与本刊稿酬一并支付㊂作者向本刊提交文章发表的行为即视为同意我刊上述声明㊂㊃81㊃。
AI预测蛋白质结构,哈萨比斯认为这一切只是开始
AI预测蛋白质结构,哈萨比斯认为这一切只是开始作者：暂无来源：《世界科学》 2021年第2期编译夏冰借助人工智能，或许能揭开有关生命运作方式的部分奥秘。
” 戴密斯· 哈萨比斯（Demis Hassabis）是人工智能公司DeepMind的联合创始人。
他领衔的团队近期解决了生物学的一大艰深问题。
比赛前的媒体造势阶段，李世石很是自信。
他即将和一位特殊的选手对弈——人工智能围棋程序AlphaGo。
不过，身为人类顶尖棋手之一的李世石认为自己会轻松获胜。
“我觉得，AlphaGo的基本机制是概率计算，只不过是一台机器而已。
”李世石当时这么说。
即便是在输掉第一场比赛之后，他也认为这只是因为AlphaGo不会犯错。
接着，在第二场对局之中，这个人工智能程序下出了如今已是声名在外的第37手。
AlphaGo就像改写了围棋规则一样，下出了根本不会有人类想到的一手棋。
最终以1：4的总比分输掉比赛的李世石惊呆了：“这手棋实在是太有想象力、太巧妙了。
”李世石和AlphaGo的这个五番棋比赛是人工智能发展历程中的重要时刻，而德米斯· 哈萨比斯正是推动人工智能发展的主要人物之一。
2010年，他与合作者共同创建了DeepMind公司，并且开始研发比人类围棋水平更高的人工智能。
那个时候，人们普遍认为，人工智能很难精通围棋，毕竟这项棋类运动中可能出现的变化要比宇宙中的原子总数还多。
然而，AlphaGo 在2016年战胜了李世石，这也让DeepMind——以及哈萨比斯——迅速成为全世界的焦点。
1997年，IBM公司的“深蓝”击败了国际象棋世界冠军加里· 卡斯帕罗夫（Garry Kasparov）。
AlphaGo在2016年的这场胜利是继深蓝之后，人工智能的又一大重要时刻。
2016年之后，DeepMind——现在，哈萨比斯的这家公司已经隶属于谷歌的母公司Alphabet 了——一直在完善算法，并且把目光投向了棋类项目以外。
Predicting Transition in Turbomachinery
ABSTRACT Here we report on an effort to include an empirically based transition modeling capability in a RANS solver. Testing of wellknown empirical models from literature for both attached- and separated-flow transition against cascade data revealed that the models did not provide enough fidelity for implementation in an airfoil design system. Consequently, a program was launched to develop models that would provide sufficient accuracy for use in an airfoil design system. As a first step in the effort, accurate modeling of freestream turbulence development was identified as a need for any form of transition modeling capability. Additionally, capturing the effects of freestream turbulence on pre-transitional boundary layers was found to have a significant effect on the accuracy of transition modeling. A CFD-supplemented database of experimental cascade cases (57 with attached-flow transition and 47 with separation and turbulent reattachment) was constructed to explore the development of new correlations. Dimensional analyses were performed to guide the work and appropriate non-dimensional parameters were then extracted from CFD predictions of the laminar boundary layers existing on the airfoil surfaces prior to either transition onset or incipient separation. For attached-flow transition, exploration of the database revealed a distinct correlation between local levels of freestream turbulence intensity, turbulence length scale, and momentum-thickness Reynolds number at transition onset. It was found that the correlation could be recast as a ratio of the boundary-layer diffusion time to a time-scale associated with the energy-bearing turbulent eddies. In the case of separatedflow transition, it was found that the length of a separation bubble prior to turbulent re-attachment was a simple function of the local momentum thickness at separation and the overall surface length traversed by a fluid element prior to separation. Both the attachedand separated-flow transition models were implemented into the design system as point-like trips. INTRODUCTION In axial-flow turbomachinery, the design trend is toward increasing airfoil loading in an effort to reduce weight and cost of future systems. Transition prediction is critical for accurate loss predictions of high lift airfoils, and the full multi-moded (Mayle, 1991) nature of the transition process must be considered. Lakshminarayana (1991), Simoneau and Simon (1993), Simon and Ashpis (1996), Dunn (2001), and Yaras (2002) all provide detailed reviews of the state of
ChatGPT技术的推理能力与常识推理技巧
ChatGPT技术的推理能力与常识推理技巧ChatGPT 是 OpenAI 开发的一种人工智能对话模型，它通过大规模训练来提供人类般的对话能力。
最近发布的 GPT-3 版本取得了非常显著的进展，其具备了惊人的语言生成能力，可以回答一系列复杂问题，并参与具有深度的交互对话。
然而，尽管其在语言生成方面表现出色，但面临的一个挑战是缺乏推理能力和常识推理技巧。
推理是一种基于已知信息进行新信息推导的能力。
它可以帮助我们从不完整的信息中进行推测和总结，并产生合理的结论。
与人类相比，ChatGPT 在推理能力方面仍然有所欠缺。
这主要是因为 ChatGPT 的训练是通过监督式学习来完成的，其输出是基于训练集中的输入和输出样本。
这种训练方法限制了 ChatGPT 对于推理能力的发展，因为推理需要更高级别的思维和复杂的模式识别。
但是，尽管 ChatGPT 的推理能力相对较弱，它仍然可以通过一些技巧进行常识推理。
常识推理是基于普遍知识和经验进行的推理。
通过利用 ChatGPT 的大规模训练数据，我们可以辅助模型进行常识推理的训练。
首先，我们可以使用大规模的预训练数据集来为 ChatGPT 提供大量的常识信息。
将 ChatGPT 暴露于各种各样的对话中可以帮助它获取不同领域的知识，从而增加其常识推理的能力。
这使得 ChatGPT 能够更好地理解问题，并提供更准确的答案。
其次，我们可以使用知识图谱等结构化的常识知识源来辅助 ChatGPT 进行常识推理。
知识图谱是一种将实体、关系和属性组织起来的图结构，可以表示各种常识知识。
通过将知识图谱与 ChatGPT 相结合，我们可以为模型提供更具体和准确的常识信息，从而提升其在推理中的表现。
除了这些技巧，我们还可以引入外部推理引擎来提升 ChatGPT 的推理能力。
这些推理引擎可以通过逻辑推理、求解问题和推断出结论等技术来扩展 ChatGPT的推理能力。
将 ChatGPT 与这些引擎相结合，可以充分发挥它们各自的优势，实现更高水平的推理。
【视野】有一种融合叫学以致用
【视野】有一种融合叫学以致用2017-09-06 17:57密歇根州立大学华盛顿大学学前教育早期干预博士（Ph.D）国际应用行为分析师（BCBA）密歇根州立大学博士后Hegarty Fellow美国早期干预协会评估组成员（上图是Project SEARCH官方网站首页）青少年特殊人士求职培训项目Project SEARCH在1996年，Erin那时还是美国辛辛那提儿童医院急诊科的主管。
她有一天突然有个想法，既然咱们医院擅长对发育障碍儿童及青少年的干预康复，那为什么咱们不能带个头，去首先聘请那些经过康复训练后有能力的青少年特殊人士呢？哪怕是让他们先从事如后勤管理之类流动率高但工作任务相对简单的职位也好啊？于是她把这个想法和同事分享，并联合社区的特殊人士就业转介中心，发起了Project SEARCH项目，专门为发育障碍等特殊人士提供实习培训和就业机会。
大学校园实习（Project SEARCH实习生在密歇根州立大学进行培训）第一天来到密歇根州立大学的校园，Spartan Project SEARCH的实习生Lisa感到既兴奋又紧张，一方面这个服务4万多学生的校园对她来说实在是太大了，单是如何坐校内巴士或者安全过行人道就已经很有挑战——她自己家的那条街一天不会有多于3辆汽车经过。
另一方面她又觉得很幸福，穿着印有校徽的衣服，和其他大学生同在一个饭堂吃饭，她感觉自己就是学校的一份子。
Lisa是唐氏综合症患者，和她一届的10个实习生有自闭症大脑发育障碍。
但是经过三期轮岗实习培训，她当过办公室行政，小卖部售货员，以及动物护理员。
她发现自己原来也有能力学习新的技能，并且为周围人贡献自己的一分力量。
实习期结束后，有两个部门都给她发了正式招聘邀请。
和家人商量后，她决定做一名动物护理员。
密歇根州立大学是密歇根州内参与该项目的12个试点单位之一。
Nick是该项目的执行经理，在9月初，他会联合康复治疗系的Connie Sung等教授，给新一届学生提供系统的实习前培训，涉及的内容包括自我管理，社交沟通等。