Warning: Declaration of action_plugin_tablewidth::register(&$controller) should be compatible with DokuWiki_Action_Plugin::register(Doku_Event_Handler$controller) in /s/bach/b/class/cs545/public_html/fall16/lib/plugins/tablewidth/action.php on line 93
code:perceptron [CS545 fall 2016]

# CS545 fall 2016

### Sidebar

CS545

Instructor
Asa Ben-Hur

code:perceptron

#### The perceptron

Here's the code for the perceptron classifier we discussed in class:

perceptron.py
import numpy as np
from matplotlib import pyplot as plt

class Perceptron :

"""An implementation of the perceptron algorithm.
Note that this implementation does not include a bias term"""

def __init__(self, max_iterations=100, learning_rate=0.2) :

self.max_iterations = max_iterations
self.learning_rate = learning_rate

def fit(self, X, y) :
"""
Train a classifier using the perceptron training algorithm.
After training the attribute 'w' will contain the perceptron weight vector.

Parameters
----------

X : ndarray, shape (num_examples, n_features)
Training data.

y : ndarray, shape (n_examples,)
Array of labels.

"""
self.w = np.zeros(len(X[0]))
converged = False
iterations = 0
while (not converged and iterations < self.max_iterations) :
converged = True
for i in range(len(X)) :
if y[i] * self.discriminant(X[i]) <= 0 :
self.w = self.w + y[i] * self.learning_rate * X[i]
converged = False
plot_data(X, y, self.w)
iterations += 1
self.converged = converged
if converged :
print ('converged in %d iterations ' % iterations)

def discriminant(self, x) :
return np.inner(self.w, x)

def predict(self, X) :
"""
make predictions using a trained linear classifier

Parameters
----------

X : ndarray, shape (num_examples, n_features)
Training data.
"""

scores = np.inner(self.w, X)
return np.sign(scores)

def generate_separable_data(N) :
w = np.random.uniform(-1, 1, 2)
print (w,w.shape)
X = np.random.uniform(-1, 1, [N, 2])
print (X,X.shape)
y = np.sign(np.dot(X, w))
return X,y,w

def plot_data(X, y, w) :
fig = plt.figure(figsize=(5,5))
plt.xlim(-1,1)
plt.ylim(-1,1)
a = -w[0]/w[1]
pts = np.linspace(-1,1)
plt.plot(pts, a*pts, 'k-')
cols = {1: 'r', -1: 'b'}
for i in range(len(X)):
plt.plot(X[i][0], X[i][1], cols[y[i]]+'o')
plt.show()

if __name__=='__main__' :
X,y,w = generate_separable_data(40)
p = Perceptron()
p.fit(X,y)