python深度学习感知器(jupyter note)

In [13]:

# 原文https://www.zybuluo.com/hanbingtao/note/433855
# 2019.5 code adapted to python 3.x
# perception 感知器
# input: [x1,x2,x3]
# weight: [w1,w2,w3]
# bias: b
# Layer1_result = input * weight + b 
# Activitaion Function 达到阈值才可以激活
# 使用AF(L1)一次次地检查每轮输入与权重 （对于and or，输入是有限的4行，且不变；变化的是权重
# ,对于and可以学习得到weight_list_a，对于or，可以学习得到weight_list_o）
# 重要概念(1): 已知的每行结果，叫做label。
# 重要概念(2): 调整weight的幅度，叫做学习速率。
# 重要概念(3)：back propagation，根据 计算结果与label的差距 * 学习速率 * 输入值 得到下一次的权重。

Output:

In [29]:

# map(f,list_a,list_b)函数: 
#第一步，定义一个函数f或者lambda函数；
#第二步，对于list a中的每一个元素，执行f or lambda；得到个数不变的新列表。
list_a = [1,3,5]
print(list_a)
iter_b = map(lambda x: x+1, list_a) #python 3 return a literable
print(list(iter_b))
print(list(map(lambda x, w: x * w, [1, 3, 5, 7, 9], [2, 4, 6, 8, 10])))

Output:

[1, 3, 5]
[2, 4, 6]
[2, 12, 30, 56, 90]

In [39]:

# reduce(f,only_one_list , initial value)函数：
# 假设累加列表[1,2,3,4] 
# 第一步： 1+2 = 3
# 第二步： 3+3 = 6
# 以此类推，直到 累计值 + 4
# ((1+2)+3) + 4）
from functools import reduce
list_r = [1,2,3,4]
t = reduce(lambda x,y : x+y, list_r, 0.0) #return is not iterable
print(t)

Output:

10.0

In [65]:

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,...] 对应在一起
            # 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
            return self.activator(
                reduce(lambda a, b: a + b,map(lambda x,w: x * w,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)
                print("%d iteration done!" % (i+1))
                print(10*"=")
        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)
            samples = zip(input_vecs, labels)
            # 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
                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,...] 对应在一起
            # 变成[(x1,w1),(x2,w2),(x3,w3),...]
            # 然后利用感知器规则更新权重
            delta = label - output
            self.weights = list(map(lambda x, w: w + rate * delta * x,input_vec, self.weights))
            print("intermediate weights:")
            print(self.weights)
            # 更新bias
            self.bias += rate * delta

Output:

In [66]:

# exampe 1 : 使用这个通用的感知器，训练出 and 函数。
# 这里的输入x:label是最简单的2:1，复杂的N:1可以是非常大的比例。
def f(x):
        '''
        定义激活函数f
        '''
        return 1 if x > 0 else 0  #这里采用的是个阶跃函数，是激活函数的一种
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]))

Output:

intermediate weights:
[0.1, 0.1]
intermediate weights:
[0.1, 0.1]
intermediate weights:
[0.0, 0.1]
intermediate weights:
[0.0, 0.1]
1 iteration done!
==========
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.0, 0.2]
intermediate weights:
[0.0, 0.1]
2 iteration done!
==========
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.1]
3 iteration done!
==========
intermediate weights:
[0.2, 0.2]
intermediate weights:
[0.2, 0.2]
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.2]
4 iteration done!
==========
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.2]
5 iteration done!
==========
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.2]
6 iteration done!
==========
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.2]
7 iteration done!
==========
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.2]
8 iteration done!
==========
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.1, 0.2]
intermediate weights:
[0.