In [6]:

    

%matplotlib inline
import numpy as np
import matplotlib.pyplot as plt
from IPython.html.widgets import interact

from sklearn.datasets import load_digits
digits = load_digits()
print(digits.data.shape)


    

    




(1797, 64)






    




:0: FutureWarning: IPython widgets are experimental and may change in the future.

In [ ]:

    

weights = np.random.randn(65)
inputs = digits.data[0:1200]
answers = digits.target
lc = 0.02


    

In [10]:

    

class Perceptron(object):
    def __init__(self, inputs, answers, weights, lc):
        self.inputs = inputs
        aelf.answers = answers
        self.weights = weights
        self.lc = lc
    
    def activate(self, sum_):
        if sum_ > 0:
            return 1
        else:
            return -1

    def sum_(self, inputs):
        sum_ = 0
        for n in range(len(weights)):
            sum_ += inputs[n]*weights[n]
        return activate(sum_)

    def train(self, inputs, answer):
        guess = sum_(inputs)
        error = answer - guess
        for n in range(len(weights)):
            weights[n] += lc*error*inputs[n]


    

In [ ]:

    

def sigmoid(z):
    return 1/(1 + exp(-z))

def sig_prime(z):
    return sigmoid(z) * (1 - sigmoid(z))