深度学习算法与编程 (暂停更新)

深度学习算法与编程

前言

本书内容

• 内容提要
  – https://blog.csdn.net/oBrightLamp/article/details/85162129

资料推荐

• 自学深度学习之计算机视觉的入门资料推荐
  – https://blog.csdn.net/oBrightLamp/article/details/84076410

开源许可 LICENSE

所有的说明性文档基于 Creative Commons 协议, 所有的代码基于 MIT 协议.
All documents are licensed under the Creative Commons License, all codes are licensed under the MIT License.

软件版本

Python = 3.6
scikit-learn = 0.20.0
TensorFlow = 1.12
PyTorch = 1.0

损失函数

MSELoss

• 均方差损失函数MSELoss详解及反向传播中的梯度求导
  – https://blog.csdn.net/oBrightLamp/article/details/85137756

cross-entropy

• 通过函数图像介绍信息熵的概念
  – https://blog.csdn.net/oBrightLamp/article/details/85269091

• 案例详解cross-entropy交叉熵损失函数及反向传播
  – https://blog.csdn.net/oBrightLamp/article/details/83962147

• Python和PyTorch对比实现cross-entropy交叉熵损失函数及反向传播
  – https://blog.csdn.net/oBrightLamp/article/details/84029058

softmax

• softmax函数详解及误差反向传播的梯度求导
  – https://blog.csdn.net/oBrightLamp/article/details/83959185

• 纯Python和PyTorch对比实现softmax及其反向传播
  – https://blog.csdn.net/oBrightLamp/article/details/84034658

softmax + cross-entropy

• 多标签softmax + cross-entropy交叉熵损失函数详解及反向传播中的梯度求导
  – https://blog.csdn.net/oBrightLamp/article/details/84069835

• Python和PyTorch对比实现多标签softmax + cross-entropy交叉熵损失及反向传播
  – https://blog.csdn.net/oBrightLamp/article/details/84073485

优化算法

正则化 / 参数规范惩罚

• L2正则化Regularization详解及反向传播的梯度求导
  – https://blog.csdn.net/oBrightLamp/article/details/85290929

SGD, Momentum, RMSprop, Adam

• 常用梯度下降算法SGD, Momentum, RMSprop, Adam详解
  – https://blog.csdn.net/oBrightLamp/article/details/85218783

• 纯Python和PyTorch对比实现SGD, Momentum, RMSprop, Adam梯度下降算法
  – https://blog.csdn.net/oBrightLamp/article/details/85218799

Batch Normalization

• Batch Normalization函数详解及反向传播中的梯度求导
  – https://blog.csdn.net/oBrightLamp/article/details/84332455

• Python和PyTorch对比实现批标准化Batch Normalization函数及反向传播
  – https://blog.csdn.net/oBrightLamp/article/details/84557854

• Batch Normalization的测试或推理过程及样本参数更新方法
  – https://blog.csdn.net/oBrightLamp/article/details/85391056

• Python和PyTorch对比实现批标准化 Batch Normalization 函数在测试或推理过程中的算法
  – https://blog.csdn.net/oBrightLamp/article/details/85391167

特征工程

PCA

• 特征工程PCA降维方法的最大方差理论详解
  – https://blog.csdn.net/oBrightLamp/article/details/85255895

• 纯Python和scikit-learn对比实现PCA特征降维
  – https://blog.csdn.net/oBrightLamp/article/details/85255898

全连接神经网络

Affine

• affine/linear(仿射/线性)变换函数详解及全连接层反向传播的梯度求导
  – https://blog.csdn.net/oBrightLamp/article/details/84333111

• Python和PyTorch对比实现affine/linear(仿射/线性)变换函数及全连接层的反向传播
  – https://blog.csdn.net/oBrightLamp/article/details/84453996

卷积神经网络

convolution

• 卷积convolution函数详解及反向传播中的梯度求导
  – https://blog.csdn.net/oBrightLamp/article/details/84561088

• Python和PyTorch对比实现卷积convolution函数及反向传播
  – https://blog.csdn.net/oBrightLamp/article/details/84589545

• 卷积convolution函数的矩阵化计算方法及其梯度的反向传播
  – https://blog.csdn.net/oBrightLamp/article/details/85870773

• Python 实现 TensorFlow 和 PyTorch 验证卷积 convolution 函数矩阵化计算及反向传播
  – https://blog.csdn.net/oBrightLamp/article/details/85870813

Transpose Convolution

• TensorFlow和PyTorch对比理解卷积和反向卷积或转置卷积(Transpose Convolution)
  – https://blog.csdn.net/oBrightLamp/article/details/85708124

MaxPool

• 池化层MaxPool函数详解及反向传播的公式推导
  – https://blog.csdn.net/oBrightLamp/article/details/84635346

• Python和PyTorch对比实现池化层MaxPool函数及反向传播
  – https://blog.csdn.net/oBrightLamp/article/details/84635308

ReLU

• ReLU函数详解及反向传播中的梯度求导
  – https://blog.csdn.net/oBrightLamp/article/details/84326978

• Python和PyTorch对比实现ReLU函数及反向传播
  – https://blog.csdn.net/oBrightLamp/article/details/84326804

dropout

• dropout函数详解及反向传播中的梯度求导
  – https://blog.csdn.net/oBrightLamp/article/details/84105097

• Python和PyTorch对比实现dropout函数及反向传播
  – https://blog.csdn.net/oBrightLamp/article/details/84326091

循环神经网络

RNN

• 循环神经网络RNNCell单元详解及反向传播的梯度求导
  – https://blog.csdn.net/oBrightLamp/article/details/85015325

• 纯Python和PyTorch对比实现循环神经网络RNNCell及反向传播
  – https://blog.csdn.net/oBrightLamp/article/details/85015402

• 纯Python和PyTorch对比实现循环神经网络RNN及反向传播
  – https://blog.csdn.net/oBrightLamp/article/details/85015387

LSTM

• 长短期记忆网络LSTMCell单元详解及反向传播的梯度求导
  – https://blog.csdn.net/oBrightLamp/article/details/85068285

• 纯Python和PyTorch对比实现循环神经网络LSTM及反向传播
  – https://blog.csdn.net/oBrightLamp/article/details/85069255

GRU

• 门控循环单元GRUCell详解及反向传播的梯度求导
  – https://blog.csdn.net/oBrightLamp/article/details/85109589

• 纯Python和PyTorch对比实现门控循环单元GRU及反向传播
  – https://blog.csdn.net/oBrightLamp/article/details/85109607

生成对抗网络

GAN

• 生成对抗网络 GAN 的数学原理
  – https://blog.csdn.net/oBrightLamp/article/details/86553074

实战

Kaggle

• PyTorch Kaggle 快速上手(杂草幼苗图片识别)
  – https://blog.csdn.net/oBrightLamp/article/details/84947499

附录

• 常用Latex公式
  – https://blog.csdn.net/oBrightLamp/article/details/83964331