I am building a neural network to learn to recognize handwritten digits from MNIST. I have confirmed that backpropagation calculates the gradients perfectly (gradient checking gives error < 10 ^ -10).

It appears that no matter how I train the weights, the cost function always tends towards around 3.24-3.25 (never below that, just approaching from above) and the training/test set accuracy is very low (around 11% for the test set). It appears that the h values in the end are all very close to 0.1 and to each other.

I cannot find why my program cannot produce better results. I was wondering if anyone could maybe take a look at my code and please tell me any reasons for this occurring. Thank you so much for all your help, I really appreciate it!

Here is my Python code:

import numpy as np
import math
from tensorflow.examples.tutorials.mnist import input_data

# Neural network has four layers
# The input layer has 784 nodes
# The two hidden layers each have 5 nodes
# The output layer has 10 nodes
num_layer = 4
num_node = [784,5,5,10]
num_output_node = 10

# 30000 training sets are used
# 10000 test sets are used
# Can be adjusted
Ntrain = 30000
Ntest = 10000

# Sigmoid Function
def g(X):
    return 1/(1 + np.exp(-X))

# Forwardpropagation
def h(W,X):
    a = X
    for l in range(num_layer - 1):
        a = np.insert(a,0,1)
        z = np.dot(a,W[l])
        a = g(z)
    return a      

# Cost Function
def J(y, W, X, Lambda):
    cost = 0
    for i in range(Ntrain):
        H = h(W,X[i])
        for k in range(num_output_node):            
            cost = cost + y[i][k] * math.log(H[k]) + (1-y[i][k]) * math.log(1-H[k])
    regularization = 0
    for l in range(num_layer - 1):
        for i in range(num_node[l]):
            for j in range(num_node[l+1]):
                regularization = regularization + W[l][i+1][j] ** 2
    return (-1/Ntrain * cost + Lambda / (2*Ntrain) * regularization)

# Backpropagation - confirmed to be correct
# Algorithm based on https://www.coursera.org/learn/machine-learning/lecture/1z9WW/backpropagation-algorithm
# Returns D, the value of the gradient
def BackPropagation(y, W, X, Lambda):
    delta = np.empty(num_layer-1, dtype = object)
    for l in range(num_layer - 1):
        delta[l] = np.zeros((num_node[l]+1,num_node[l+1]))
    for i in range(Ntrain):
        A = np.empty(num_layer-1, dtype = object)
        a = X[i]
        for l in range(num_layer - 1):
            A[l] = a
            a = np.insert(a,0,1)
            z = np.dot(a,W[l])
            a = g(z)
        diff = a - y[i]
        delta[num_layer-2] = delta[num_layer-2] + np.outer(np.insert(A[num_layer-2],0,1),diff)
        for l in range(num_layer-2):
            index = num_layer-2-l
            diff = np.multiply(np.dot(np.array([W[index][k+1] for k in range(num_node[index])]), diff), np.multiply(A[index], 1-A[index])) 
            delta[index-1] = delta[index-1] + np.outer(np.insert(A[index-1],0,1),diff)
    D = np.empty(num_layer-1, dtype = object)
    for l in range(num_layer - 1):
        D[l] = np.zeros((num_node[l]+1,num_node[l+1]))
    for l in range(num_layer-1):
        for i in range(num_node[l]+1):
            if i == 0:
                for j in range(num_node[l+1]):
                    D[l][i][j] = 1/Ntrain * delta[l][i][j]
                for j in range(num_node[l+1]):
                    D[l][i][j] = 1/Ntrain * (delta[l][i][j] + Lambda * W[l][i][j])
    return D

# Neural network - this is where the learning/adjusting of weights occur
# W is the weights
# learn is the learning rate
# iterations is the number of iterations we pass over the training set
# Lambda is the regularization parameter
def NeuralNetwork(y, X, learn, iterations, Lambda):

    W = np.empty(num_layer-1, dtype = object)
    for l in range(num_layer - 1):
        W[l] = np.random.rand(num_node[l]+1,num_node[l+1])/100
    for k in range(iterations):
        print(J(y, W, X, Lambda))
        D = BackPropagation(y, W, X, Lambda)
        for l in range(num_layer-1):
            W[l] = W[l] - learn * D[l]
    print(J(y, W, X, Lambda))
    return W

mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)

# Training data, read from MNIST
inputpix = []
output = []

for i in range(Ntrain):
    inputpix.append(2 * np.array(mnist.train.images[i]) - 1)

np.savetxt('input.txt', inputpix, delimiter=' ')
np.savetxt('output.txt', output, delimiter=' ')

# Train the weights
finalweights = NeuralNetwork(output, inputpix, 2, 5, 1)

# Test data
inputtestpix = []
outputtest = []

for i in range(Ntest):
    inputtestpix.append(2 * np.array(mnist.test.images[i]) - 1)

np.savetxt('inputtest.txt', inputtestpix, delimiter=' ')
np.savetxt('outputtest.txt', outputtest, delimiter=' ')

# Determine the accuracy of the training data
count = 0
for i in range(Ntrain):
    H = h(finalweights,inputpix[i])
    for j in range(num_output_node):
        if H[j] == np.amax(H) and output[i][j] == 1:
            count = count + 1

# Determine the accuracy of the test data
count = 0
for i in range(Ntest):
    H = h(finalweights,inputtestpix[i])
    for j in range(num_output_node):
        if H[j] == np.amax(H) and outputtest[i][j] == 1:
            count = count + 1

2 Answers 2


If you are sure that the code for forward and backward passes is correct, then it seems that the problem is with model's hyperparameters as well as in using only 30k images.

If accuracy is ~11%, is means that the model always predicts one and the same value.

  1. Learning rate is 2. This is very high. Usually it is less than 0.1, or even 10-100 smaller.
  2. Number of iterations is 5. While this could be ok, it seems low.
  3. Regularization strength is 1. Quite high, try 0.1-0.001
  4. Hidden layers. There are 2 hidden layers with 5 nodes. 5 nodes is quite low. While there are a lot of theories about the number of neurons in hidden layers, it shouldn't be lower than the number of outputs. Try at least 10, or maybe better 64 or 128.
  5. You use 30000 training samples. Is your PC not powerfull enough or is there any other reason? Anyway, the problem is that classes maybe disbalanced in these 30000, some classes may have too much samples and som - too few. If you take only 30000 samples, you need to make sure that the classes are almost equally distributed.
  • $\begingroup$ To add to @Andrey Lukyanenko. I would advise you to study learning curves, bias and variance issues. $\endgroup$
    – Ehsan
    Commented Aug 5, 2017 at 16:09

To add to @Andrey, you should be able to achieve an accuracy of 98+% accuracy on the training with a simple 2 layer nn with 250-350 hidden nodes, sigmoid activation, learning rate of 0.1. I ran my test 1,000,000 iterations via stochastic gradient descent. However, I'm very sure it converged far earlier.

An example showing how I structured the experiment and results: here


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.