【机器学习】手写感知机模型

数据集来自mnist数据集，主要利用numpy里的matrix矩阵计算。算法的实现主要在于对参数w和b的求解。算法的推导过程参考李航《统计学习方法》，推导最优化函数然后更新参数的过程。

import numpy as np
import pandas as pd
import time

def data_load(filename):
    '''
    :param filename:
    :return: dataArr,labelArr
    '''
    print('start read file')
    dataArr,labelArr = [],[]
    with open(filename,'r') as f:
        lines = f.readlines()
    for line in lines:
        line = line.strip().split(',')

        if int(line[0]) >= 5:
            labelArr.append(1)
        else:
            labelArr.append(-1)

        dataArr.append([int(num)/255 for num in line[1:]])
    print('End')
    return dataArr,labelArr

def perception(dataArr,labelArr,iter=50,lr=0.001):
    dataMat = np.mat(dataArr)
    m,n = dataMat.shape
    labelMat = np.mat(labelArr).T
    w = np.zeros((1,n))
    b = 0
    # loss = 0
    for k in range(iter):
        loss = 0
        for i in range(len(dataArr)):
            x_i = dataMat[i]
            y_i = labelMat[i]
            if y_i * (w * x_i.T + b) <= 0:
                w = w + lr * y_i * x_i
                b = b + lr * y_i

            new_y = w * x_i.T + b
            loss += abs(new_y - y_i)
        print('iter:{}'.format(k))
    return w,b

def data_test(dataArr,labelArr,w,b):
    dataMat = np.mat(dataArr)
    labelMat = np.mat(labelArr).T
    sum = len(dataArr)
    rigSum = 0
    for i in range(len(dataArr)):
        x_i = dataMat[i]
        y_i = labelMat[i]
        result = -1 * y_i * (w * x_i.T + b)
        if result <= 0:
            rigSum += 1
    return rigSum/sum*100



if __name__ == '__main__':
    dataArr, labelArr = data_load('../dataset/mnist_train.csv')
    start = time.time()
    w,b = perception(dataArr,labelArr)
    end = time.time()
    print(data_test(dataArr,labelArr,w,b))
    print(end - start)