In [ ]:

    

import matplotlib as mpl
mpl.use('Agg')
%matplotlib inline
import matplotlib.pyplot as plt
import numpy as np
import os
import pandas as pd
import seaborn as sns
import sys


    

In [ ]:

    

sys.path.append('../code')


    

In [ ]:

    

from NeuralNet import NeuralNet
from TransferFunctions import TanhTF, LinearTF


    

In [ ]:

    

from sklearn.datasets import make_classification
X, y = make_classification(n_samples=2000, n_features=10, 
                           n_informative=10, n_redundant=0, n_repeated=0, 
                           n_classes=2, n_clusters_per_class=1, 
                           weights=None, flip_y=0.0001, class_sep=1.0, 
                           hypercube=True, shift=0.0, scale=1.0, 
                           shuffle=True, random_state=None)
X=X.T
split_frac = 0.25
N = X.shape[1]
split_index = int(N*(1-split_frac))
print("{} of {} points from training are reserved for "
      "validation".format(N - split_index, N))
test_X = X[:, split_index:]
test_y = y[split_index:]
X = X[:, 0:split_index]
y = y[0:split_index]

print('shape of X: {}.  (Columns are rows; rows are features)'.format(X.shape))


    

In [ ]:

    

y.shape


    

In [ ]:

    

# columns are data points and rows are features
d, N = np.shape(X)
C = np.unique(y).shape[0]


    

In [ ]:

    

n = NeuralNet(X=X, y=y, hidden_nodes=20, 
              hiddenTF=TanhTF, outputTF=TanhTF,
              #hiddenTF=LinearTF, outputTF=LinearTF,
              minibatch_size=10,
              eta0 = 0.001,
              X_test = test_X, y_test = test_y
              )


    

In [ ]:

    

n.N


    

In [ ]:

    

n.run(epochs=20)


    

In [ ]:

    

n.C


    

In [ ]:

    

n.results.tail()


    

In [ ]:

    

sl = n.plot_square_loss()


    

In [ ]:

    

l01 = n.plot_01_loss()


    

In [ ]:

    

p = n.plot_weight_evolution()


    

In [ ]:

    

p1 = n.plot_sum_of_weights('W1')
p2 = n.plot_sum_of_weights('W2')


    

In [ ]:

    

p1 = n.plot_norm_of_gradient('W1')
p2 = n.plot_norm_of_gradient('W2')


    

In [ ]:

    

n.W1_dot_prod_checking.head()


    

In [ ]:

    

n.W2_dot_prod_checking.head()


    

In [ ]:

    

p = n.plot_sample_dot_prods() #figsize=(8,3))


    

In [ ]:

    

np.linalg.norm(np.array([[1, 2]]))


    

In [ ]:

    

n.predict_y(n.X)


    

In [ ]:

    

dW1, dW2 = n.gradients(n.X, n.Y, rounded=True)


    

In [ ]:

    

dW1_num, dW2_num = n.numerical_derivatives()


    

In [ ]:

    

np.max(dW1 - dW1_num)


    

In [ ]:

    

np.sum(np.abs(dW1 - dW1_num))


    

In [ ]:

    

np.sum(np.abs(dW2 - dW2_num))


    

In [ ]:

    

n.results.tail(1).T


    

In [ ]: