神经网络英语演讲
相关主题
- 1、下载文档前请自行甄别文档内容的完整性，平台不提供额外的编辑、内容补充、找答案等附加服务。
- 2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
- 3、如文档侵犯您的权益，请联系客服反馈,我们会尽快为您处理(人工客服工作时间：9:00-18:30)。
1.4 ELM(extreme learning machine)
• Zero • One unified
form • generalized
2 Deep NNRWs(neural network with random weights)
2.1 AE(auto-encoder)+NNRWs
• 3-layer
1.2 perception
• Retina of sensory units • A set of association cells • Response units
1.3 RVFL(random vector functional link networks)
• [-1,1]
• MooreCPenrose pseudo-inverse.
• 2 Deep neural network with random weights.
• 3 Conclusion.
1 Shallow feed-forword neural network with random weights
1.1 Turing
• The father of AI (artificial intelligence) • Turing Test • Can Machines Think?
English Presentation
Title
•A review on neural networks with random weights.
Contents
• 1 Shallow feed-forword neural network with random weights(NNRWs).
• Disadvantages: The randomization range. The type of distribution of the hidden weights.
Cao W, Wang X, Ming Z, et al. A Review on Neural Networks with Random Weights[J]. Neurocomputing, 2017.
• Unsupervised learning
2.2 RBM(restricted boltzmann machinHale Waihona Puke Baidus)+NNRWs
• Two layers
• Fully connected
• No connections in the same layer
3 conclusion
• Advantages: Faster learning speed Improve the computing efficiency.
Thank you