Nguồn ngoài
Neural Networks and Deep Learning
Neural networks are one of the most beautiful programming paradigms ever invented. In the conventional approach to programming, we tell the computer what to do, breaking big problems up into many small, precisely defined tasks that the computer can easily perform. By contrast, in a neural network we don't tell the computer how to solve our problem. Instead, it learns from observational data, figuring out its own solution to the problem at hand.
- Tác giả
- Michael Nielson
- Nhà xuất bản
- Y Combinator Research
- Nguồn
- LibreTexts
- Giấy phép
- CC BY-NC
- Ngôn ngữ
- EN
- Danh mục
- Công nghệ thông tin
Mục lục· 39 mục
- 01: Using neural nets to recognize handwritten digits
- 1.01: Introduction
- 1.02: Perceptrons
- 1.03: Sigmoid neurons
- 1.04: The architecture of neural networks
- 1.05: A simple network to classify handwritten digits
- 1.06: Learning with gradient descent
- 1.07: Implementing our network to classify digits
- 1.08: Toward deep learning
- 02: How the Backpropagation Algorithm Works
- 2.01: Introduction
- 2.02: Warm up- a fast matrix-based approach to computing the output from a neural network
- 2.03: The backpropagation algorithm
- 03: Improving the way neural networks learn
- 3.01: The cross-entropy cost function
- 3.02: Overfitting and regularization
- 3.03: Weight initialization
- 3.04: Handwriting recognition revisited- the code
- 3.05: How to choose a neural network's hyper-parameters
- 3.06: Other techniques
- 04: A visual proof that neural nets can compute any function
- 4.01: Two caveats
- 4.02: Universality with one input and one output
- 4.03: Many input variables
- 4.04: Extension beyond sigmoid neurons
- 4.05: Fixing up the step functions
- 4.06: Section 6-
- 05: Why are deep neural networks hard to train
- 5.01: The vanishing gradient problem
- 5.02: Unstable gradients in more complex networks
- 06: Deep Learning
- 6.01: Introducing convolutional networks
- 6.02: Convolutional neural networks in practice
- 6.03: The code for our convolutional networks
- 6.04: Recent progress in image recognition
- 6.05: Other approaches to deep neural nets
- 6.06: On the future of neural networks
- 07: Appendix- Is there a simple algorithm for intelligence
- 7.01: Is there a simple algorithm for intelligence