1. 池化和采样PoolingMax/Avg poolingupsamplenearestbilinearRelu2. cifar10013 layersCodeimport tensorflow as tffrom tensorflow.keras import layers, optimizers, metrics, dat...

1. 如何通俗易懂地解释卷积？转载自知乎，马同学2. 2D Convolution3. Padding & Stride4. Channels5. layers.Conv2D6. 需要全套课程视频+PPT+代码资源可以私聊我方式1：优快云私信我!方式2：QQ邮箱:594042358@qq.com或者直接加我QQ: 594042358!...

1. Keras.Metrics （度量指标）1.1. Build a meter1.2. Update data1.3. Get Average data1.4. Clear buffer1.5. Codeimport tensorflow as tffrom tensorflow.keras import datasets, layers, optimizers, Se...

1. 梯度下降（Gradient Descent）1.1. What’s Gradient1.2. What does it mean1.3. How to Search1.4. AutoGradGradientTapePersistent GradientTape2nd-orderimport tensorflow as tfw = tf.Var...

1. 均方误差MSE2. 交叉熵损失Cross Entropy Loss2.1. 信息熵/信息增益2.2. Cross Entropy2.3. Classification2.4. Why not MSE？2.5. svm损失函数Hinge Loss3. logits→CrossEntropy4. Numerical Stabilityimport tenso...

1. Outline1.1. Matmul （矩阵形式）???????????? = ????(????@???? + ????)???????????? = ????????????????(????@???? + ????)X@W+b1.2. Neural Network （神经网络结构）lecun在86年89年在手写数字论文中用到的就是这个网络， 当时叫做神经网络。3~5 layers1.3. Deep Learning （深度学习）n 约等于 ...

1. tensorflow.keras.datasets1.1. 常用数据集boston housing （波士顿的房价）Boston housing price regression dataset.mnist/fashion mnist （手写数字）MNIST/Fashion-MNIST dataset.cifar10/100 （小型的图片识别数据集）smal...

123

1. 前向传播案例1.1. 代码实现import tensorflow as tffrom tensorflow import kerasfrom tensorflow.keras import datasets# 加载数据集# x: [60k, 28, 28]# y: [60k] 0~9(x, y), _ = datasets.mnist.load_data()# x:[...

1. Outline±*/**, pow, squaresqrt//, %exp, log@, matmullinear layer2. Operation typeelement-wise±*/matrix-wise@, matmuldim-wisereduce_mean/max/min/sum2.1. ±*/%//2.2. tf.ma...

1. Broadcasting1.1. Outlineexpandwithout copying dataVS tf.tiletf.broadcast_to1.2. Key ideaInsert 1 dim ahead if neededExpand dims with size 1 to same sizeFeature maps: [4, 32, 32, 3]B...

1. Outlineshape, ndimreshapeexpand_dims/squeezetranspose2. Reshape2.1. Reshape is flexible2.2. Reshape could lead to potential bugs!3. tf.transpose4. Squeeze VS Expand_dims4.1. Expand ...

1. Indexing1.1. Basic indexing1.2. Same With Numpy1.2.1. Numpy-style indexing1.2.2. start:end1.2.3. Indexing by :1.2.4. Indexing by ::1.2.5. ::-11.2.6. …

1. 数据类型1.1. 数据容器listnp.arraytf.Tensor1.2. What’s Tensorscalar: 1.1vector: [1.1],[1.1, 2.2, … ]matrix: [[1.1, 2.2],[3.3, 4.4],[5.5, 6.6]]tensor: ???????????????? > 2 (在神经网络中， tensor的范围相当广， 这里可以理解为狭义...

1. 手写数字识别简述1.1. MNIST7000 images per categorytrain/test splitting: 60k vs 10k1.2. Image[28,28,1]→[784]1.3. Input and Output1.4. Regression VS Classification???? = ???? ∗ ???? + ???????? ∈ ????????????????????...

1. 线性回归1.1. Continuous Prediction1.2. Linear Equation1.3. With Noise?1.4. For Example1.5. Find ????′, ????′1.6. Gradient Descent2. Numpy实现线性回归2.1 Find ????′, ????′2.2. SolutionCompute LossComp...

1. 发展历史1.1. 前世今生2015.9发布0.1版本2017.2发布1.0版本2019春发布2.0版本1.2. 同一时期人工智能其他库的发展1.2.1. 2015年Scikit-learnMachine learning, No GPUCaffe2013, 第一个面向深度学习的框架No auto-grad, C++KeraswrapperT...