PERCEPTRON

import numpy as np

input = np.array([
                  [-1,-1],
                  [-1,1],
                  [1,-1],
                  [1,1]
])
target=np.array([-1,1,1,1])
w=[0,0]
LR=1
b=0
epoch= 20
T=0

for j in range (0, epoch):
  print("epoch:",j)
  global_error = 0
  for i in range(0,input.shape[0]):

    actual = target[i]
    instance = input[i]

    x0 = instance[0]
    x1 = instance[1]

    sum_unit = b + (x0*w[0] + x1*w[1])

    if sum_unit > 0:
      activation=1
    elif sum_unit < 0:
      activation=-1
    else:
      activation =0

    if actual != activation:
      w[0]=w[0] + LR*actual*x0
      w[1]=w[1] + LR*actual*x1
      b=b + LR*actual

    error = actual - activation
    print("prediction:", activation,"actual:",target[i],"error:",error,")")

    global_error = global_error +abs(error)
    '''
    error= actual-activation
    global_error = global_error+abs(error)
    print("prediction:",activation,"whereas actual was:", target[i],"(error:",error.")")
    w[0]=w[0]+LR*error
    W[1]=w[1]+LR*error
    #print(w[0])
    #print(w[1])
    #b=b+actual*LR
  print("---------------")

  if global error==0:
    break

    '''

  print("-----------------")
  
  if global_error ==0:
    print("final weight:",w[0],w[1],b)

    break