In this handson lecture we will implement a 3-layer neural network from scratch.
Note: this example is adapted from this blog post and notebook to make it easier to work through without seeing the solution. It is well worth reading through the original post and notebook as well, but we advise you avoid peeking during this class exercise!
In [1]:
# Package imports
import matplotlib.pyplot as plt
import numpy as np
import sklearn
import sklearn.datasets
import sklearn.linear_model
import matplotlib
# Display plots inline and change default figure size
%matplotlib inline
matplotlib.rcParams['figure.figsize'] = (10.0, 8.0)
In [2]:
# Generate a dataset and plot it
np.random.seed(0)
X, y = sklearn.datasets.make_moons(200, noise=0.20)
plt.scatter(X[:,0], X[:,1], s=40, c=y, cmap=plt.cm.Spectral)
Out[2]:
In [3]:
num_examples = len(X) # training set size
nn_input_dim = 2 # input layer dimensionality
nn_output_dim = 2 # output layer dimensionality
# Gradient descent parameters (I picked these by hand)
epsilon = 0.01 # learning rate for gradient descent
reg_lambda = 0.01 # regularization strength
Let's implement functions for different pieces of the puzzle for backprop. We'll hold our model in a dictionary with the following keys:
Note: the methods below reference the dataset variables X and y directly as free parameters (global to this notebook).
In [4]:
# Helper function to predict an output (0 or 1)
def predict(model, x):
W1, b1, W2, b2 = model['W1'], model['b1'], model['W2'], model['b2']
# Forward propagation
z1 = x.dot(W1) + b1
a1 = np.tanh(z1)
z2 = a1.dot(W2) + b2
exp_scores = np.exp(z2)
probs = exp_scores / np.sum(exp_scores, axis=1, keepdims=True)
return np.argmax(probs, axis=1)
In [5]:
# Helper function to evaluate the total loss on the dataset
def calculate_loss(model):
W1, b1, W2, b2 = model['W1'], model['b1'], model['W2'], model['b2']
# Forward propagation to calculate our predictions
z1 = X.dot(W1) + b1
a1 = np.tanh(z1)
z2 = a1.dot(W2) + b2
exp_scores = np.exp(z2)
probs = exp_scores / np.sum(exp_scores, axis=1, keepdims=True)
# Calculating the loss
corect_logprobs = -np.log(probs[range(num_examples), y])
data_loss = np.sum(corect_logprobs)
# Add regulatization term to loss (optional)
data_loss += reg_lambda/2 * (np.sum(np.square(W1)) + np.sum(np.square(W2)))
return 1./num_examples * data_loss
The loss function is $$L(y, \hat{y}) = -\frac{1}{N}\sum_{n\in N}\sum_{i\in C}y_{n, i}\log \hat{y}_{n, i}$$.
To train the parameters of our network, we'll need to use backprop to compute the following gradients at each step:
$$ \delta_3 = y - \hat{y} \\\ \delta_2 = (1 - \tanh^2 z_1)\delta_3 W_2^T \\\ \frac{\partial L}{\partial w_2} = a_1^T \delta_3 \\\ \frac{\partial L}{\partial b_2} = \delta_3 \\\ \frac{\partial L}{\partial W_1} = x^T \delta_2 \\\ \frac{\partial L}{\partial b_1} = \delta2 $$The code below builds a model and trains it. We've left out the backpropogation part for you to fill in.
Hints:
a1 which already computes $\tanh z_1$
In [6]:
# This function learns parameters for the neural network and returns the model.
# - nn_hdim: Number of nodes in the hidden layer
# - num_passes: Number of passes through the training data for gradient descent
# - print_loss: If True, print the loss every 1000 iterations
def build_model(nn_hdim, num_passes=20000, print_loss=False):
# Initialize the parameters to random values. We need to learn these.
np.random.seed(0)
W1 = np.random.randn(nn_input_dim, nn_hdim) / np.sqrt(nn_input_dim)
b1 = np.zeros((1, nn_hdim))
W2 = np.random.randn(nn_hdim, nn_output_dim) / np.sqrt(nn_hdim)
b2 = np.zeros((1, nn_output_dim))
# This is what we return at the end
model = {}
# Gradient descent. For each batch...
for i in range(0, num_passes):
# Forward propagation
z1 = X.dot(W1) + b1
a1 = np.tanh(z1)
z2 = a1.dot(W2) + b2
exp_scores = np.exp(z2)
probs = exp_scores / np.sum(exp_scores, axis=1, keepdims=True)
# Backpropagation
delta3 = probs
delta3[range(num_examples), y] -= 1
# update the code below to compute dW2, db2, delta2, dW1 and db1
dW2 = 0
db2 = 0
delta2 = 0
dW1 = 0
db1 = 0
# Add regularization terms (b1 and b2 don't have regularization terms)
dW2 += reg_lambda * W2
dW1 += reg_lambda * W1
# Gradient descent parameter update
W1 += -epsilon * dW1
b1 += -epsilon * db1
W2 += -epsilon * dW2
b2 += -epsilon * db2
# Assign new parameters to the model
model = {'W1': W1, 'b1': b1, 'W2': W2, 'b2': b2}
# Optionally print the loss.
# This is expensive because it uses the whole dataset, so we don't want to do it too often.
if print_loss and i % 1000 == 0:
print("Loss after iteration %i: %f" % (i, calculate_loss(model)))
return model
In [7]:
# Build a model with a 3-dimensional hidden layer
model = build_model(3, num_passes=8000, print_loss=True)
In [8]:
# Helper function to plot a decision boundary.
# If you don't fully understand this function don't worry, it just generates the contour plot below.
def plot_decision_boundary(pred_func):
# Set min and max values and give it some padding
x_min, x_max = X[:, 0].min() - .5, X[:, 0].max() + .5
y_min, y_max = X[:, 1].min() - .5, X[:, 1].max() + .5
h = 0.01
# Generate a grid of points with distance h between them
xx, yy = np.meshgrid(np.arange(x_min, x_max, h), np.arange(y_min, y_max, h))
# Predict the function value for the whole gid
Z = pred_func(np.c_[xx.ravel(), yy.ravel()])
Z = Z.reshape(xx.shape)
# Plot the contour and training examples
plt.contourf(xx, yy, Z, cmap=plt.cm.Spectral)
plt.scatter(X[:, 0], X[:, 1], c=y, cmap=plt.cm.Spectral)
In [9]:
# Plot the decision boundary
plot_decision_boundary(lambda x: predict(model, x))
plt.title("Decision Boundary for hidden layer size 3")
Out[9]: