descente de gradient dans le plateau d'entraînement du réseau neuronal

J'ai essayé d'implémenter un réseau neuronal de base de rétro-propogation en python et j'ai terminé la programmation pour l'initialisation et l'entraînement de l'ensemble de poids. Cependant, sur tous les ensembles que j'entraîne, l'erreur (moyenne au carré) converge toujours vers un nombre étrange - l'erreur diminue toujours lors des itérations ultérieures, mais ne s'approche jamais vraiment de zéro.
Toute aide serait très appréciée.

import csv
import numpy as np

class NeuralNetwork:
layers = 0
shape = None
weights = []

layerIn = []
layerOut = []

def __init__(self, shape):
    self.shape = shape
    self.layers = len(shape) - 1

    for i in range(0,self.layers):
        n = shape[i]
        m = shape[i+1]
        self.weights.append(np.random.normal(scale=0.2, size = (m,n+1)))

def sgm(self, x):
    return 1/(1+np.exp(-x))

def dersgm(self, x):
    y = self.sgm(x)
    return y*(y-1)


def run(self, input):
    self.layerIn = []
    self.layerOut = []

    for i in range(self.layers):
        if i == 0:
            layer = self.weights[0].dot(np.vstack((input.transpose(), np.ones([1,input.shape[0]]))))
        else:
            layer = self.weights[i].dot(np.vstack((self.layerOut[-1], np.ones([1,input.shape[0]]))))
        self.layerIn.append(layer)
        self.layerOut.append(self.sgm(layer))

    return self.layerOut[-1].T

def backpropogate(self, input, y, learning_rate):
    deltas = []
    y_hat = self.run(input)

    #Calculate deltas
    for i in reversed(range(self.layers)):

        #for last layer
        if i == self.layers-1:
            error = y_hat - y
            msq_error = sum(.5 * ((error) ** 2))
            #returns delta, k rows for k inputs, m columns for m nodes
            deltas.append(error * self.dersgm(y_hat))
        else:

            error = deltas[-1].dot(self.weights[i+1][:,:-1])
            deltas.append(self.dersgm(self.layerOut[i]).T * error)

    #Calculate weight-deltas
    wdelta = []
    ordered_deltas = list(reversed(deltas)) #reverse order because created backwards

    #returns weight deltas, k rows for k nodes, m columns for m next layer nodes
    for i in range(self.layers):
        if i == 0:
            #add bias
            input_with_bias = np.vstack((input.T, np.ones(input.shape[0])))
            #some over n rows of deltas for n training examples to get one delta for all examples
            #for all nodes
            wdelta.append(ordered_deltas[i].T.dot(input_with_bias.T))
        else:
            with_bias = np.vstack((self.layerOut[i-1], np.ones(input.shape[0])))
            wdelta.append(ordered_deltas[i].T.dot(with_bias.T))



    #update_weights
    def update_weights(self, weight_deltas, learning_rate):
        for i in range(self.layers):
            self.weights[i] = self.weights[i] +\
                              (learning_rate * weight_deltas[i])


    update_weights(self, wdelta, learning_rate)

    return msq_error

    #end backpropogate

def train(self, input, target, lr, run_iter):
    for i in range(run_iter):
        if i % 100000 == 0:
            print self.backpropogate(input, target, lr)