原文地址 Almost any process you can imagine can be thought of as function computation. Of course, just because we know a neural network exists that can (say) translate Chinese text into English, tha...
Neural Networks And Deep Learning Chap6
原文地址 Introducing convolutional networks Local receptive fields 局部采样 shared weights 分享权重 To see why this makes sense, suppose the weights and bias are such that the hidden neuron can p...
Neural Networks And Deep Learning Chap5
原文地址 Why are deep neural networks hard to train? 前一章有介绍，任意函数都可以通过2层神经网络来求解，但是并不是说任何问题都用两层就好了。因为“可以做到”并不一定是“最好的”。 作者的观点是，当使用2层神经网络来解决问题的时候，需要的神经元的数量可能是多层网络的指数倍，设计的困难程度也比多层复杂得多。 Deep circuits thu...
Neural Networks And Deep Learning Chap3
原文地址 a better choice of cost function, known as the cross-entropy cost function four so-called “regularization” methods (L1 and L2 regularization, dropout, and artificial expansion of the tra...
Neural Networks And Deep Learning Chap2
原文地址 Warm up: a fast matrix-based approach to computing the output from a neural network The two assumptions we need about the cost function 两个假设： \[C=\frac{1}{2n}\sum_x \left \| y(x)-a^{L}(...
Neural Networks And Deep Learning Chap1
原文地址 perceptrons（感知器） 1950s-1960s by scientist Frank Rosenblatt 数学模型： 所有权重参数为w1,w2…，threshold threshold，Dropping the threshold means you’re more willing to go to the festival. 与权重b是同一种意思，...