import math
import random
import string
random.seed(0)
# 生成區(qū)間[a, b)內(nèi)的隨機(jī)數(shù)
def rand(a, b):
 return (b-a)*random.random() + a
# 生成大小 I*J 的矩陣，默認(rèn)零矩陣 (當(dāng)然，亦可用 NumPy 提速)
def makeMatrix(I, J, fill=0.0):
 m = []
 for i in range(I):
  m.append([fill]*J)
 return m
# 函數(shù) sigmoid，這里采用 tanh，因?yàn)榭雌饋?lái)要比標(biāo)準(zhǔn)的 1/(1+e^-x) 漂亮些
def sigmoid(x):
 return math.tanh(x)
# 函數(shù) sigmoid 的派生函數(shù), 為了得到輸出 (即：y)
def dsigmoid(y):
 return 1.0 - y**2
class NN:
 ''' 三層反向傳播神經(jīng)網(wǎng)絡(luò) '''
 def __init__(self, ni, nh, no):
  # 輸入層、隱藏層、輸出層的節(jié)點(diǎn)（數(shù)）
  self.ni = ni + 1 # 增加一個(gè)偏差節(jié)點(diǎn)
  self.nh = nh
  self.no = no
  # 激活神經(jīng)網(wǎng)絡(luò)的所有節(jié)點(diǎn)（向量）
  self.ai = [1.0]*self.ni
  self.ah = [1.0]*self.nh
  self.ao = [1.0]*self.no
  # 建立權(quán)重（矩陣）
  self.wi = makeMatrix(self.ni, self.nh)
  self.wo = makeMatrix(self.nh, self.no)
  # 設(shè)為隨機(jī)值
  for i in range(self.ni):
   for j in range(self.nh):
    self.wi[i][j] = rand(-0.2, 0.2)
  for j in range(self.nh):
   for k in range(self.no):
    self.wo[j][k] = rand(-2.0, 2.0)
  # 最后建立動(dòng)量因子（矩陣）
  self.ci = makeMatrix(self.ni, self.nh)
  self.co = makeMatrix(self.nh, self.no)
 def update(self, inputs):
  if len(inputs) != self.ni-1:
   raise ValueError('與輸入層節(jié)點(diǎn)數(shù)不符！')
  # 激活輸入層
  for i in range(self.ni-1):
   #self.ai[i] = sigmoid(inputs[i])
   self.ai[i] = inputs[i]
  # 激活隱藏層
  for j in range(self.nh):
   sum = 0.0
   for i in range(self.ni):
    sum = sum + self.ai[i] * self.wi[i][j]
   self.ah[j] = sigmoid(sum)
  # 激活輸出層
  for k in range(self.no):
   sum = 0.0
   for j in range(self.nh):
    sum = sum + self.ah[j] * self.wo[j][k]
   self.ao[k] = sigmoid(sum)
  return self.ao[:]
 def backPropagate(self, targets, N, M):
  ''' 反向傳播 '''
  if len(targets) != self.no:
   raise ValueError('與輸出層節(jié)點(diǎn)數(shù)不符！')
  # 計(jì)算輸出層的誤差
  output_deltas = [0.0] * self.no
  for k in range(self.no):
   error = targets[k]-self.ao[k]
   output_deltas[k] = dsigmoid(self.ao[k]) * error
  # 計(jì)算隱藏層的誤差
  hidden_deltas = [0.0] * self.nh
  for j in range(self.nh):
   error = 0.0
   for k in range(self.no):
    error = error + output_deltas[k]*self.wo[j][k]
   hidden_deltas[j] = dsigmoid(self.ah[j]) * error
  # 更新輸出層權(quán)重
  for j in range(self.nh):
   for k in range(self.no):
    change = output_deltas[k]*self.ah[j]
    self.wo[j][k] = self.wo[j][k] + N*change + M*self.co[j][k]
    self.co[j][k] = change
    #print(N*change, M*self.co[j][k])
  # 更新輸入層權(quán)重
  for i in range(self.ni):
   for j in range(self.nh):
    change = hidden_deltas[j]*self.ai[i]
    self.wi[i][j] = self.wi[i][j] + N*change + M*self.ci[i][j]
    self.ci[i][j] = change
  # 計(jì)算誤差
  error = 0.0
  for k in range(len(targets)):
   error = error + 0.5*(targets[k]-self.ao[k])**2
  return error
 def test(self, patterns):
  for p in patterns:
   print(p[0], '->', self.update(p[0]))
 def weights(self):
  print('輸入層權(quán)重:')
  for i in range(self.ni):
   print(self.wi[i])
  print()
  print('輸出層權(quán)重:')
  for j in range(self.nh):
   print(self.wo[j])
 def train(self, patterns, iterations=1000, N=0.5, M=0.1):
  # N: 學(xué)習(xí)速率(learning rate)
  # M: 動(dòng)量因子(momentum factor)
  for i in range(iterations):
   error = 0.0
   for p in patterns:
    inputs = p[0]
    targets = p[1]
    self.update(inputs)
    error = error + self.backPropagate(targets, N, M)
   if i % 100 == 0:
    print('誤差 %-.5f' % error)
def demo():
 # 一個(gè)演示：教神經(jīng)網(wǎng)絡(luò)學(xué)習(xí)邏輯異或（XOR）------------可以換成你自己的數(shù)據(jù)試試
 pat = [
  [[0,0], [0]],
  [[0,1], [1]],
  [[1,0], [1]],
  [[1,1], [0]]
 ]
 # 創(chuàng)建一個(gè)神經(jīng)網(wǎng)絡(luò)：輸入層有兩個(gè)節(jié)點(diǎn)、隱藏層有兩個(gè)節(jié)點(diǎn)、輸出層有一個(gè)節(jié)點(diǎn)
 n = NN(2, 2, 1)
 # 用一些模式訓(xùn)練它
 n.train(pat)
 # 測(cè)試訓(xùn)練的成果（不要吃驚哦）
 n.test(pat)
 # 看看訓(xùn)練好的權(quán)重（當(dāng)然可以考慮把訓(xùn)練好的權(quán)重持久化）
 #n.weights()
if __name__ == '__main__':
 demo()