目录概念反向传播算法CNNLeNetAlexNetVGGRNNLSTM 概念 感知机, perceptron, 多层感知机（multilayer perceptron，MLP） Sigmoid函数，即 f(x)=1/(1+e−x)f(x)=1/(1+e^{-x})f(x)=1/(1+e−x) Softplus函数，f(x)=log(1+ex)f(x) = log(1 + e^x)f(x)=log(1+ex) 修正线性单元（Rectified linear unit，ReLU） 前馈神经网络（feedforward neural networks） 递归神经网络（recurrent neura...

目录神经网络浅层神经网络activation functionsigmoidtanhReLULeaky ReLU随机初始化He initializationXavier initialization正则化L2范数DropoutData augmentationEarly stopping归一化优化算法指数移动加权平均MomentumRMSpropAdamAdaGradNesterov超参数调优Batch NormalizationSoftMaxML工程化CNN边缘检测PaddingStride三维卷积Pooling(下采样)原理反向传播经典网络LeNet-5AlexNetVGG-16ResNe...

目录基础5-fold cross validationApproximate Nearest Neighbor (ANN)t-SNENCARandom Projection¶L2 distance向量化方法Multi-class SVMHOGColor HistogramBag of wordsCNNYOLO/SDDMask R-CNNsemantic/instance segmentation可解释DeepDreamGANAutoencoderGANReinforce LearningQ LearningRecurrent Atention model 基础 5-fold cross val...