Function main

cnn_class/benchmark.py:63–142

()

def main():
    train, test = get_data()
    

    # Need to scale! don't leave as 0..255
    # Y is a N x 1 matrix with values 1..10 (MATLAB indexes by 1)
    # So flatten it and make it 0..9
    # Also need indicator matrix for cost calculation
    Xtrain = flatten(train['X'].astype(np.float32) / 255.)
    Ytrain = train['y'].flatten() - 1
    Xtrain, Ytrain = shuffle(Xtrain, Ytrain)

    Xtest  = flatten(test['X'].astype(np.float32) / 255.)
    Ytest  = test['y'].flatten() - 1

    # gradient descent params
    max_iter = 20
    print_period = 10
    N, D = Xtrain.shape
    batch_sz = 500
    n_batches = N // batch_sz

    # initial weights
    M1 = 1000 # hidden layer size
    M2 = 500
    K = 10
    W1_init = np.random.randn(D, M1) / np.sqrt(D + M1)
    b1_init = np.zeros(M1)
    W2_init = np.random.randn(M1, M2) / np.sqrt(M1 + M2)
    b2_init = np.zeros(M2)
    W3_init = np.random.randn(M2, K) / np.sqrt(M2 + K)
    b3_init = np.zeros(K)

    # define variables and expressions
    X = tf.placeholder(tf.float32, shape=(None, D), name='X')
    T = tf.placeholder(tf.int32, shape=(None,), name='T')
    W1 = tf.Variable(W1_init.astype(np.float32))
    b1 = tf.Variable(b1_init.astype(np.float32))
    W2 = tf.Variable(W2_init.astype(np.float32))
    b2 = tf.Variable(b2_init.astype(np.float32))
    W3 = tf.Variable(W3_init.astype(np.float32))
    b3 = tf.Variable(b3_init.astype(np.float32))

    Z1 = tf.nn.relu( tf.matmul(X, W1) + b1 )
    Z2 = tf.nn.relu( tf.matmul(Z1, W2) + b2 )
    logits = tf.matmul(Z2, W3) + b3

    cost = tf.reduce_sum(
        tf.nn.sparse_softmax_cross_entropy_with_logits(
            logits=logits,
            labels=T
        )
    )

    train_op = tf.train.RMSPropOptimizer(0.0001, decay=0.99, momentum=0.9).minimize(cost)

    # we'll use this to calculate the error rate
    predict_op = tf.argmax(logits, 1)