动手实现感知器

样例程序：

from functools import reduce

class Perceptron(object):
    def __init__(self, input_num, activator):
        ‘‘‘
        初始化感知器，设置输入参数的个数，以及激活函数。
        激活函数的类型为double -> double
        ‘‘‘
        self.activator = activator
        # 权重向量初始化为0
        self.weights = [0.0 for _ in range(input_num)]
        # 偏置项初始化为0
        self.bias = 0.0
    def __str__(self):
        ‘‘‘
        打印学习到的权重、偏置项
        ‘‘‘
        return ‘weights\t:%s\nbias\t:%f\n‘ % (self.weights, self.bias)
    def predict(self, input_vec):
        ‘‘‘
        输入向量，输出感知器的计算结果
        ‘‘‘
        # 把input_vec[x1,x2,x3...]和weights[w1,w2,w3,...]打包在一起
        # 变成[(x1,w1),(x2,w2),(x3,w3),...]
        # 然后利用map函数计算[x1*w1, x2*w2, x3*w3]
        # 最后利用reduce求和
        print("predict")
        print(input_vec)
        print(self.weights)
        print(zip(input_vec, self.weights))
        return self.activator(
            reduce(lambda a, b: a + b,
                   map(lambda x_w: x_w[0] * x_w[1],
                       zip(input_vec, self.weights))
                , 0.0) + self.bias)
    def train(self, input_vecs, labels, iteration, rate):
        ‘‘‘
        输入训练数据：一组向量、与每个向量对应的label；以及训练轮数、学习率
        ‘‘‘
        for i in range(iteration):
            self._one_iteration(input_vecs, labels, rate)
    def _one_iteration(self, input_vecs, labels, rate):
        ‘‘‘
        一次迭代，把所有的训练数据过一遍
        ‘‘‘
        # 把输入和输出打包在一起，成为样本的列表[(input_vec, label), ...]
        # 而每个训练样本是(input_vec, label)
        print("_one_iteration")
        print(input_vecs)
        print(labels)
        samples = zip(input_vecs, labels)
        print(samples)
        # 对每个样本，按照感知器规则更新权重
        for (input_vec, label) in samples:
            # 计算感知器在当前权重下的输出
            print(input_vec)
            output = self.predict(input_vec)
            # 更新权重
            self._update_weights(input_vec, output, label, rate)
    def _update_weights(self, input_vec, output, label, rate):
        ‘‘‘
        按照感知器规则更新权重
        ‘‘‘
        # 把input_vec[x1,x2,x3,...]和weights[w1,w2,w3,...]打包在一起
        # 变成[(x1,w1),(x2,w2),(x3,w3),...]
        # 然后利用感知器规则更新权重
        delta = label - output
        #print("_update_weights")
        #print(input_vec)
        #print(self.weights)
        #print(rate)
        #print(delta)
        #for each in zip(input_vec, self.weights)
            #print(each)
        #for a,b in zip(input_vec, self.weights):
            #print(‘a: %f‘ % a)
            #print(‘b: %f‘ % b)
        self.weights = [w + rate * delta * x for x,w in zip(input_vec, self.weights)]
        # 更新bias
        self.bias += rate * delta

继续：

def f(x):
    ‘‘‘
    定义激活函数f
    ‘‘‘
    return 1 if x > 0 else 0
def get_training_dataset():
    ‘‘‘
    基于and真值表构建训练数据
    ‘‘‘
    # 构建训练数据
    # 输入向量列表
    input_vecs = [[1,1], [0,0], [1,0], [0,1]]
    # 期望的输出列表，注意要与输入一一对应
    # [1,1] -> 1, [0,0] -> 0, [1,0] -> 0, [0,1] -> 0
    labels = [1, 0, 0, 0]
    return input_vecs, labels
def train_and_perceptron():
    ‘‘‘
    使用and真值表训练感知器
    ‘‘‘
    # 创建感知器，输入参数个数为2（因为and是二元函数），激活函数为f
    p = Perceptron(2, f)
    # 训练，迭代10轮, 学习速率为0.1
    input_vecs, labels = get_training_dataset()
    p.train(input_vecs, labels, 10, 0.1)
    #返回训练好的感知器
    return p
if __name__ == ‘__main__‘:
    # 训练and感知器
    and_perception = train_and_perceptron()
    # 打印训练获得的权重
    print(and_perception)
    # 测试
    print(‘1 and 1 = %d‘ % and_perception.predict([1, 1]))
    print(‘0 and 0 = %d‘ % and_perception.predict([0, 0]))
    print(‘1 and 0 = %d‘ % and_perception.predict([1, 0]))
    print(‘0 and 1 = %d‘ % and_perception.predict([0, 1]))

自己实现的部分：

class Perceptron(object):#8

    def __init__(self):#9
        self.w = [0,0]
        self.b = [0]

    def get(self):#10
        return self.w,self.b

    def getw(self):#13
        return self.w

    def getb(self):#14
        return self.b

    def setw(self,x):#15
        self.w = x

    def setb(self,x):#16
        self.b[0] = x

    def activator(self,x):#12
        if x>0:
            return 1
        else:
            return 0

    def predict(self,x):#11
        results = self.activator(self.w[0]*x[0]+self.w[1]*x[1]+self.b[0])
        return results

继续：

def produce():#2
    #inputs=([0,0],[0,1],[1,0],[1,1])
    #labels=(0,0,0,1)
    inputs=[[1,1],[0,0],[0,1],[1,0],[1,1]]
    labels=[1,0,0,0,1]
    print(inputs)
    print(labels)
    return inputs,labels

‘‘‘
def init():#3
    w = [0,0]
    b = [0]
    print(w)
    print(b)
    return w,b
‘‘‘

def deprecated():#4
    inputs,labels = produce()
    w,b = init()
    #results = map(lambda x: w*x,inputs)
    ‘‘‘
    print(inputs)
    print(w)
    for each in zip(inputs,w)
        print(each[0])
        print(each[1])
    results = map(lambda x: x[0]*x[1],zip(inputs,w))
    print(results)
    ‘‘‘
    #print([0,1]*[1,1])

    ‘‘‘
def activator(x):#5
    if x>0:
        return x
    else:
        return 0
    ‘‘‘

    ‘‘‘
def predict():#6
    inputs,labels = produce()
    w,b = init()
    results = []
    for each in inputs:
        #print(type(w[0]))
        #print(type(each[0]))
        results.append(activator(w[0]*each[0]+w[1]*each[1]+b[0]))
    #print(results)
    return results
    ‘‘‘

‘‘‘
def predict(x):#7
    w,b = init()
    return results = activator(w[0]*x[0]+w[1]*x[1]+b[0])
‘‘‘

def one_iteration(inputs,p):#17
    for i in range(len(inputs)):
        #print(inputs[i])
        #print("labels: %d" % labels[i])
        delta = labels[i]-p.predict(inputs[i])
        delta_w = map(lambda x: 0.1 * delta * x, inputs[i]) #0.1 * delta * inputs[i]
        delta_b = 0.1 * delta
        #print(list(delta_w))
        delta_w = list(delta_w)
        #print(delta_b)
        #delta_b = delta_b[0]#list(delta_b)
        w_tmp=[]
        w_tmp.append(p.getw()[0]+delta_w[0])
        w_tmp.append(p.getw()[1]+delta_w[1])
        p.setw(w_tmp)
        p.setb(p.getb()[0]+delta_b)

def train(inputs,p,epoch):#18
    for i in range(epoch):
        one_iteration(inputs,p)
        #print("iteration:%d" % i)
        #print(p.getw())
        #print("b:%f" % p.getb()[0])

if __name__ == ‘__main__‘:#1
    #results = predict()
    inputs,labels = produce()
    p = Perceptron()
    epoch = 1
    train(inputs,p,epoch)
    #print(p.getw())
    #print(p.getb())
    print("1 and 1 :%d" % p.predict([1,1]))
    #delta = map(lambda x:x[1]-x[0],zip(results,labels))
    #print(list(delta))

输出：

[[1, 1], [0, 0], [0, 1], [1, 0], [1, 1]]
[1, 0, 0, 0, 1]
1 and 1 :1

参考：

https://www.cnblogs.com/ratels/p/11427328.html

原文地址：https://www.cnblogs.com/ratels/p/11432655.html