Pytorch深度学习入门与实战一--全连接神经网络

全连接神经网络在分类和回归问题中都非常有效，本节对全连接神经网及神经元结构进行介绍，并使用Pytorch针对分类和回归问题分别介绍全连接神经网络的搭建，训练及可视化相关方面的程序实现。
1.全连接神经网络简介

import numpy as np
import pandas as pd
from sklearn.preprocessing import StandardScaler,MinMaxScaler
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score,confusion_matrix,calsaaification_report
from sklearn.manifold import TSNE
import torch
import torch.nn as nn
from torch.optim import SGD,Adam
from torch.utils.data as Data
import matplotlib.pyplot as plt
import seaborn as sns
import hiddenlayer as hl
from torchviz import make_dot

**2.MLP分类

数据准备和探索
本节使用一个垃圾邮件数据介绍如何使用PyTorch建立MLP分类模型，该数据集可以从UCI机器学习数据库进行下载。

https://archive.ics.uci.edu/ml/datasets/Spambase

该数据集中包含57个邮件内容的统计特征，其中：
48个特征是关键词出现的频率x100的取值，范围为[0,100]，变量名使用word_freq_WORD命名，WORD表示该特征统计的词语。
6个特征为关键字符出现的频率x100取值，取值范围为[0,100]，变量名为char_freq_CHAR命名
1个变量为captical_run_length_average，表示大写字母不间断的平均长度
1个变量captial_run_length_total表示邮件中大写字母的数量。
数据集中最后一个变量是带预测目标变量(0,1)，表示电子邮件被认为是垃圾邮件（1）或者不是（0）

 spam=pd.read_csv("data/chap5/spambase.csv")
spam.head()
pd.value_counts(spam.label)X=spam.iloc[:,0:57].values
y=spam.label.values
X_train,X_test,y_train,y_test=train_test_split(X,y,test_size=0.25,random_state=123)scales=MinMaxScaler(featre_range=(0,1))
X_train_s=scales.fit_trandform(X_train)
X_test_s=scales.fit_trandform(X_test)colname=spam.cloumns.values[:-1]
plt.figure(figsize=(20,14))
for ii in range(len(colname)):plt.subplot(7,9,ii+1)sns.boxplot(x=y_train,y=X_train_s[:,ii])plt.title(colname[ii])
plt.subplots_adjust(hspace=0.4)
plt.show()

搭建网络并可视化

 class MLPclassifica(nn.Module):def __init__(self):super(MLPclassifica,self).__init__()self.hidden1=nn.Sequential(nn.Linear(in_features=57,out_features=30,bias=True,),nn.ReLU())self.hidden2=nn.Sequential(nn.Linear(30,10),nn.ReLU())self.classifica=nn.Sequential(nn.Linear(10,2),nn.Sigmoid())def forward(self,x):fc1=self.hidden1(x)fc2=self.hidden2(fc1)output=self.classifica(fc2)return fc1,fc2,outputmlpc=MLPclassifica()
x=torch.randn(1,57).reshapes_grad_(True)
y=mlpc(x)
Mymlpcvis=make_dot(y,params=dict(list(mlpc.named_parameters())+[('x',x)]))
Mymlpcvis

使用未预处理的数据训练模型

X_train_nots=torch.from_numpy(X_train.astype(np.float32))
y_train_t=torch.from_numpy(y_train.astype(np.int64))
X_test_nots=torch.from_numpy(X_test.astype(np.float32))
y_test_t=torch.from_numpy(y_test.astype(np.int64))
train_data_nots=Data.TensorDataset(X_train_nots,y_train_t)
train_nots_loader=Data.DataLodaer(dataset=train_data_nots,batch_size=64,shuffle=True,num_workers=1,
)
optimizer=torch.optim.Adam(mlpc.parameters(),lr=0.01)
loss_func=nn.CrossEntropyLoss()history1=h1.History()
canvas1=h1.Canvas()
print_step=25for epoch in range(15):for step,(b_x,b_y) in enumerate(train_nots_loader):_,_,output=mlpc(b_x)train_loss=loss_func(output,b_y)optimizer.zero_grad()train_loss.backward()optimizer.step()niter=epoch*len(train_nots_loader)+step+1if niter % print_step == 0:_,_,output=mlpc(X_test_nots)_,pre_lab=torch.max(output,1)test_accuracy=accuracy_score(y_test_t,pre_lab)history1.log(niter,train_loss=train_loss,test_accuracy=test_accuracy)with canvas1:canvas1.draw_plot(history1["train_loss"])canvas1.draw_plot(history1["test_accuracy"])

使用预处理的数据训练模型

X_train_t=torch.from_numpy(X_train_s.astype(np.float32))
y_train_t=torch.from_numpy(y_train.astype(np.int64))
X_test_t=torch.from_numpy(X_test_s.astype(np.float32))
y_test_t=torch.from_numpy(y_test.astype(np.int64))
train_data=Data.TensorDataset(X_train_t,y_train_t)
train_loader=Data.DataLodaer(dataset=train_data,batch_size=64,shuffle=True,num_workers=1,
)
optimizer=torch.optim.Adam(mlpc.parameters(),lr=0.01)
loss_func=nn.CrossEntropyLoss()history1=h1.History()
canvas1=h1.Canvas()
print_step=25for epoch in range(15):for step,(b_x,b_y) in enumerate(train_loader):_,_,output=mlpc(b_x)train_loss=loss_func(output,b_y)optimizer.zero_grad()train_loss.backward()optimizer.step()niter=epoch*len(train_loader)+step+1if niter % print_step == 0:_,_,output=mlpc(X_test_t)_,pre_lab=torch.max(output,1)test_accuracy=accuracy_score(y_test_t,pre_lab)history1.log(niter,train_loss=train_loss,test_accuracy=test_accuracy)with canvas1:canvas1.draw_plot(history1["train_loss"])canvas1.draw_plot(history1["test_accuracy"])_,_,output=mlpc(X_test_t)_,pre_lab=torch.max(output,1)test_accuracy=accuracy_score(y_test_t,pre_lab)print("test_accuracy:",test_accuracy)

获取中间层的输出并可视化
使用中间层的输出

 _,test_fc2,_=mlpc(X_test_t)
print("test_fc2.shape: ",test_fc2.shape)
test_fc2_tsne=TSNE(n_components=2).fit_transform(test_fc2.data.numpy())
plt.figure(figsize=(8,6))
plt.xlim([min(test_fc2_tsne[:,0]-1),max(test_fc2_tsne[:,0])+1])
plt.ylim([min(test_fc2_tsne[:,1]-1),max(test_fc2_tsne[:,1])+1])
plt.plot(test_fc2_tsne[y_test==0,0],test_fc2_tsne[y_test=0,1],"bo",label="0")
plt.plot(test_fc2_tsne[y_test==1,0],test_fc2_tsne[y_test=1,1],"rd",label="1")
plt.legend()
plt.title("test_f2_tsne")
plt.show()

使用钩子获取中间层的输出

activation={}
def get_activation(name):def hook(model,input,output):activation[name]=output.detach()return hook
mlpc.classifica.register_forward_hook(get_activation("classifica"))
_,_,_=mlpc(X_test_t)
classifica=activation["classifica"].data.numpy()
print("classifica.shape:"classifica.shape)
plt.figure(figsize=(8,6))
plt.plot(classifica[y_test==0,0],classifica[y_test=0,1],"bo",label="0")
plt.plot(classifica[y_test==1,0],classifica[y_test=1,1],"rd",label="1")
plt.legend()
plt.title("test_f2_tsne")
plt.show()

3.MLP回归模型

import numpy as np
import pandas as pd
from sklearn.preprocessing import StandardScaler,MinMaxScaler
from sklearn.model_selection import train_test_split
from sklearn.metrics import mean_squared_error,mean_absolute_error
from sklearn.datasets import fetch_california_housing
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.optim import SGD
from torch.utils.data as Data
import matplotlib.pyplot as plt
import seaborn as snshousdata=fetch_california_housing()
X_train,X_test,y_train,y_test=train_test_split(housdata.data,housdata.target,test_size=0.3,random_state=42
)
scale=StandardScaler()
X_train_s=scale.fit_transform(X_train)
X_test_s=scale.fit_transform(X_test)housedatadf=pd.DataFrame(data=X_train_s,columns=housedata.feature_names)
housedatadf["target"]=y_train
//热力图分析9个变量之间的相关性
datacor=np.corrcoef(housedatadf.values,rowvar=0)
datacor=pd.DataFrame(data=datacor,columns=housedatadf.columns,index=housedatadf.columns)
plt.figure(figsiez=(8,6))
ax=sns.heatmap(datacor,square=True,annot=True,fmt=".3f",linewidths=.5,cmap="Y1GnBu",cbar_kws={"fraction":0.046,"pad""0.03"})
plt.show()train_xt=torch.from_numpy(X_train_s.astype(np.float32))
train_yt=torch.from_numpy(y_trian.astype(np.float32))
test_xt=torch.from_numpy(X_test_s.astype(np.float32))
test_yt=torch.from_numpy(y_test.astype(np.float32))train_data=Data.TensorDataset(train_xt,train_yt)
test_data=Data.TensorDataset(test_xt,test_yt)
train_loader=Data.DataLoader(dataset=train_data,batch_size=64,shuffle=True,num_workers=1)class MPLregression(nn.Module):def __init__(self):super(MLPregression,self).__init__()self.hidden1=nn.Linear(in_features=8,out_features=100,bias=True)self.hidden2=nn.Linear(100,100)self.hidden3=nn.Linear(100,50)self.predict=nn.Linear(50,1)def forward(self,x):x=F.relu(self.hidden1(x))x=F.relu(self.hidden2(x))x=F.relu(self.hidden3(x))output=self.predict(x)return output[:,0]mlpreg=MPLregression()
print(mlpreg)optimzer=torch.optim.SGD(mlpreg.parameters(),lr=0.01)
loss_func=nn.MSELoss()
train_loss_all=[]
for epoch in range(30):train_loss=0train_num=0for step,(b_x,b_y) in enumerate(train_loader):output=mlpreg(b_x)loss=loss_func(output,b_y)optimzer.zero_grad()loss.backward()optimzer.step()train_loss+=loss.item()*b_x.size(0)train_num+=b_x.size(0)train_loss_all.append(train_loss/train_num)plt.figure(figsize=(10,6))
plt.plot(train_loss_all,"ro-",label="Train loss")
plt.legend()
plt.grid()
plt.xlabel("epoch")
plt.ylabel("Loss")
plt.show()pre_y=mlpreg(test_xt)
pre_y=pre_y.data.numpy()
mae=mean_absolute_error(y_test,pre_y)index=np.argsort(y_test)
plt.figure(figsize=(12,5))
plt.plot(np.arange(len(y_test)),y_test[index],"r",label="Original Y")
plt.scatter(np.arange(len(pre_y)),pre_y[index],s=3,c="b",label="Prediction")
plt.legend(loc="upper left")
plt.grid()
plt.xlabel("Index")
plt.ylabel("Y")
plt.show()

Pytorch深度学习入门与实战一--全连接神经网络相关推荐

PyTorch深度学习入门与实战（案例视频精讲）
作者:孙玉林,余本国著出版社:中国水利水电出版社品牌:智博尚书出版时间:2020-07-01 PyTorch深度学习入门与实战(案例视频精讲)
【深度学习】第二章：全连接神经网络
文章目录 1. 什么是全连接神经网络? 2. 三要素 2.1 模型 2.2 准则 2.3 算法 3. 反向传播算法 3.1 为什么要使用反向传播算法的随机梯度下降法求最优化? 3.2 什么是反向传播算 ...
pytorch深度学习入门与实战——今天我们来对一张图像进行卷积、池化，以及激活层的使用展示
import numpy as np import torch import torch.nn as nn import matplotlib.pyplot as plt from PIL impor ...
Pytorch深度学习入门与实战（笔记）
目录第二章 1.张量数据类型 (1)查看张量数据类型 (2)修改张量数据类型
AI深度学习入门与实战21 文本分类：用 Bert 做出一个优秀的文本分类模型
在上一讲,我们一同了解了文本分类(NLP)问题中的词向量表示,以及简单的基于 CNN 的文本分类算法 TextCNN.结合之前咱们学习的 TensorFlow 或者其他框架,相信你已经可以构建出一个属 ...
PyTorch深度学习入门
作者:曾芃壹出版社:人民邮电出版社品牌:iTuring 出版时间:2019-09-01 PyTorch深度学习入门
【金融】【pytorch】使用深度学习预测期货收盘价涨跌——全连接神经网络模型构建与训练
[金融][pytorch]使用深度学习预测期货收盘价涨跌--全连接神经网络模型构建与训练模型构建与训练模型构建与训练 def get_accuracy(SR,GT,threshold=0.5):S ...
pytorch深度学习入门笔记
Pytorch 深度学习入门笔记作者:梅如你学习来源: 公众号: 阿力阿哩哩.土堆碎念 B站视频:https://www.bilibili.com/video/BV1hE411t7RN? 中国大学 ...
Tensorflow2.0深度学习入门与实战（日月光华）（学习总结1）
Tensorflow2.0深度学习入门与实战(学习总结1) 我是刚学的,网易云课堂跟着日月光华老师,现在对每节课的学习课程做一下记录,总结,仅仅作为总结. 1.使用快捷键 shift+enter执行代 ...

Pytorch深度学习入门与实战一--全连接神经网络

Pytorch深度学习入门与实战一--全连接神经网络相关推荐

最新文章

热门文章