convex-optimization/Classifier.py at main · ko3365/convex-optimization · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
import random as rnd
import numpy as np
import cvxpy as cp
import pandas as pd
import matplotlib.pyplot as plt

class rand2DGaussian:
	def __init__(self):
		self.data = []
		self.label = []

	def add_data(self, classifier):
		self.label.append(classifier)
		if classifier == 1: #mean (-1, 1)
			self.data.append([np.random.normal(-1,1),np.random.normal(1,1)])
		elif classifier == -1: #mean (1,-1)
			self.data.append([np.random.normal(1,1),np.random.normal(-1,1)])

	#generate n random data points using add_data with labels -1 or 1
	def generate_data(self, n):
		for i in range(n):
			self.add_data(rnd.choice([-1,1]))

def isBetweenHyperplanes(x, w, b, margins):
	eq = x.dot(w) + b
	return margins[0] <= eq <= margins[1] or margins[1] <= eq <= margins[0]

#split data to training and testing
def split_data(data, labels, percent):
	train_data = []
	train_labels = []
	test_data = []
	test_labels = []
	for i in range(len(data)):
		if rnd.random() < percent:
			test_data.append(data[i])
			test_labels.append(labels[i])
		else:
			train_data.append(data[i])
			train_labels.append(labels[i])
	return np.array(train_data), np.array(train_labels), np.array(test_data), np.array(test_labels)
	# return train_data, train_labels, test_data, test_labels

def shuffle_data(data, labels):
	combined = list(zip(data, labels))
	rnd.shuffle(combined)
	data, labels = zip(*combined)
	return np.array(data), np.array(labels)

#score function for SVM
def get_score(test_data, test_labels, w, intercept):
	correct = 0
	for i in range(len(test_data)):
		if test_labels[i] == 1:
			if test_data[i].dot(w) + intercept > 0:
				correct += 1
		else:
			if test_data[i].dot(w) + intercept <= 0:
				correct += 1
	return correct/len(test_data)

class Classifier:
	def __init__(self):
		self.w = [0]*len(train_data[0])
		self.intercept = 0
		self.classes = 0
		self.selected_train_data = []
		self.selected_train_labels = []
		self.score = 0
		self.new_sample_added = False
		self.toggle_var = 1
		self.b = []
		self.b_pos = []
		self.b_neg = []
		self.regulizer = 0.05

	def sample_selection(self, training_sample):
		train_sample_data = training_sample[1:]
		train_sample_label = np.array([training_sample[0]])
		if len(self.classes) < 2:
			self.selected_train_data.append(train_sample_data)
			self.selected_train_labels.append(train_sample_label)
			self.new_sample_added = True
			self.classes = np.unique(self.selected_train_labels)
		else:
			if(self.toggle_var==train_sample_label):
				if(isBetweenHyperplanes(train_sample_data, self.w, self.intercept, self.b_neg) and train_sample_label == -1):
					self.selected_train_data.append(train_sample_data)
					self.selected_train_labels.append(train_sample_label)
					self.new_sample_added = True
					self.toggle_var = -train_sample_label
				if(isBetweenHyperplanes(train_sample_data, self.w, self.intercept, self.b_pos) and train_sample_label == 1):
					self.selected_train_data.append(train_sample_data)
					self.selected_train_labels.append(train_sample_label)
					self.new_sample_added = True
					self.toggle_var = -train_sample_label

	def train(self, train_data, train_label):
		train_data, train_label = shuffle_data(train_data, train_label)
		for i, (train_sample_data, train_sample_label) in enumerate(zip(train_data, train_label)):
			self.classes = np.unique(self.selected_train_labels)
			training_sample = np.concatenate((train_sample_label,train_sample_data))
			self.sample_selection(training_sample)

			if len(self.classes) == 2 and self.new_sample_added:
				#print('new sample', i, len(self.selected_train_data), self.new_sample_added, self.selected_train_labels[-1])
				self.new_sample_added = False
				self.selected_train_data = np.array(self.selected_train_data)
				self.selected_train_labels = np.array(self.selected_train_labels)
				Weights = cp.Variable((len(train_sample_data),1))
				gamma = cp.Variable()
				# hinge loss
				loss = cp.sum(cp.pos(1 - cp.multiply(self.selected_train_labels, self.selected_train_data @ Weights + gamma)))
				c = cp.norm(Weights, 1)
				slack = cp.Parameter(nonneg=True, value = self.regulizer)

				prob = cp.Problem(cp.Minimize(loss/len(self.selected_train_data) + c*slack))
				prob.solve()
				#get params
				self.w = Weights.value.flatten()
				# print(w)
				margin = 1/np.linalg.norm(self.w)
				#get the gab between the hyperplane and the origin
				self.intercept = gamma.value
				#get the margin of the hyperplane

				# print(margin)
				# print('w:', w, 'intercept:', intercept, 'margin:', margin)
				self.b = [float(self.intercept - 0*margin), float(self.intercept + 1*margin)]
				self.b_pos = [float(self.intercept + 0*margin), float(self.intercept + 1*margin)]
				self.b_neg = [float(self.intercept - 1*margin), float(self.intercept - 0*margin)]
				# print('b_pos:', b_pos, 'b_neg:', b_neg)
				#score the testing data

				self.selected_train_data = list(self.selected_train_data)
				self.selected_train_labels = list(self.selected_train_labels)

	def f(self, input):
		if input.dot(self.w) + self.intercept > 0:
			return 1
		else:
			return -1

	def get_score(self, test_data, test_labels):
			correct = 0
			for i in range(len(test_data)):
				if self.f(test_data[i]) == test_labels[i]:
					correct += 1
			return correct/len(test_data)

	def test(self, test_data, test_labels):
		self.score = self.get_score(test_data, test_labels)
		self.classifications = [1 if x > 0 else -1 for x in test_data @ self.w + self.intercept]
		return self.classifications

synthetic_data = rand2DGaussian()
synthetic_data.generate_data(12000)

train_data, train_labels, test_data, test_labels = split_data(synthetic_data.data, synthetic_data.label, 0.2)
#reshape train_labels
train_labels = np.reshape(train_labels, (len(train_labels),1))
test_labels = np.reshape(test_labels, (len(test_labels),1))

classifier = Classifier()
classifier.train(train_data, train_labels)
classifier.test(test_data, test_labels)
print('Size of Train Sets: ', train_data.shape[0])
print('Size of Selected Data: ', len(classifier.selected_train_data))
print("Score using", len(classifier.selected_train_data), "data points: ", classifier.score)