qq_42344132的博客

解压数据集：import osimport zipfile# os.chdir('./classify-leaves.zip')data=zipfile.ZipFile('classify-leaves.zip','r')data.extractall()

import numpy as npimport matplotlib.pyplot as pltimport torchimport copyimport torch.nn as nnfrom torchvision.datasets import FashionMNISTfrom torchvision import transformsimport torch.utils.data as Dataimport csvimport gzipimport pandas as pdi.

来源：动手学-深度学习-softmax回归1.把标签y表示为one-hot编码这点要时刻牢记，在代码中有体现。one-hot编码：类别对应的分量设置为1，其他分量设置为0。例如“猫”为（1,0,0），“狗”为（0,1,0），“鸡”为（0,0,1）2.将模型的输出视作概率o为模型的输出（没有做归一化操作）；y_hat为归一化后的概率。3.损失函数yi为独热标签向量。上面给出的是，梯度为4.代码利用cs231n的作业的softmax代码学习。就是.

目录1.PCA2. 自编码模型1.PCAfrom sklearn import datasetsfrom sklearn import decompositionimport matplotlib.pyplot as pltimport numpy as npimport seabornfrom mpl_toolkits.mplot3d import Axes3D%matplotlib notebookdef load_data(path): with.

参考cs231n作业一的softmaximport numpy as npimport matplotlib.pyplot as pltdef load_data(path): with np.load(path) as f: x_train, y_train = f['x_train'].astype('float32').reshape(60000,784), f['y_train'] x_test, y_test = f['.

import numpy as npimport matplotlib.pyplot as pltimport randomdef unpickle(file): import pickle with open(file, 'rb') as fo: dict = pickle.load(fo, encoding='bytes') return dictX_train,y_train,X_test,y_test=[],[],[],[]for i ...