README.md

Overview

In this package, we've implemented a few basic machine learning algorithms from scratch. The algorithms are implemented in Python 3.8. The algorithms are as follows:

• Single Layer Perceptron Algorithm
• Pocket Perceptron
• Adaline Algorithm with delta rule
• Linear Regression
• Logistic Regression
• Polynomial Regression
• One-vs-All and One-vs-One Classifiers

Furthermore, we've implemented a few basic algorithms useful in the cristalization of learning theory :

• Non-Linear Transformation
• Cross Validation with K-Fold
• Gradient Descent for Linear and Logistic Regression
• Regularization for Linear and Logistic Regression
• Bias and Variance (not implemented yet)
• Vapnik Chervonenkis Dimension (VC-Dimension)
• Covering and Uniform Covering Number

Finally, we also added a few basic functions to help you create dummy data and plot results of the algorithms :

• Generate Dummy Data (in 2D and 3D)
• Plot Linear Decision Boundary (in 2D and 2D)
• Plot Non-Linear Decision Boundary (in 2D)
• Plot Linear Regression (in 2D)

All the algorithms and tools are implemented in the lib folder.

These codes were made in the context of the Learning Theory course Fall 2023 at ENSIAS - University Mohammed V - Rabat, Morocco. You can find the LABS and the corresponding PDFs in the ensias_labs folder. Do not use them since they are a beta version with a lot of bugs and errors.

Authors and Contributors

• Hamza Bamohammed, Applied Mathematics & AI engineering student at ENSIAS
• Hicham Filali, Applied Mathematics & AI engineering student at ENSIAS
• Bouchra Sahri, Applied Mathematics & AI engineering student at ENSIAS
• Mohammed Nechba, Applied Mathematics & AI engineering student at ENSIAS
• Hanaa El Afia, Applied Mathematics & AI engineering student at ENSIAS
• Mohamed Mouhajir, Applied Mathematics & AI engineering student at ENSIAS

Installation

To install the package, you can use the following command :

  git clone https://github.com/HamBa-m/scinis-learn.git

Usage/Examples

Data Generation (in 2D)

from lib.data import generateData

# make the list of circles as dictionnaries with keys 'center' and 'radius' and "class"
circles = [
    {'center': [0, 0], 'radius': 1, 'class': 1},
    {'center': [2, 2], 'radius': 1, 'class': -1},
    ]

# Generate 500 points in 2D
data = generateData(500, circles, filename='data2D.csv')

PS: Don't forget to manually add the features in the first line of the resulting CSV file before proceeding in this tutorial!

Perceptron Algorithm

import numpy as np
from lib.classification import PLA
from lib.data import loadData

# Load data
data = loadData('data2D.csv', ['x0','x1','x2','y'])
X = data[:, :-1]
y = data[:, -1]

# Initialize the weights vector
w = np.zeros(X.shape[1])

# Initialize the Perceptron Algorithm
w0, t = PLA(X, y, w)

# Print the weights vector and the number of iterations
print(w0, t)

Plotting Decision Boundary (in 2D)

from lib.plots import plotLinearDecisionBoundary2D

# Plot the decision boundary
plotLinearDecisionBoundary2D(X, y, w0)

Results

iter= 1  | loss= 0.0
[ 3.       -1.513614 -2.503794] 1

Logistic Regression

from lib.classification import LogisticRegression
from lib.data import loadData
import numpy as np

# Load data
data = loadData('data2Dlogistic.csv', ['x0','x1', 'x2', 'y'])
X = data[:, :-1]
y = data[:, -1]

# Initialize the logistic regression algorithm
w0, loss = LogisticRegression(X, y)

# Print the weights vector and the number of iterations
print(w0, loss)

# Plot the decision boundary
from lib.plots import plotNonLinearDecisionBoundary2D
plotNonLinearDecisionBoundary2D(X, y, w0, 1)

Results

100%    ==================================================>[ 4.94597336 -2.8523291  -2.58496018] 0.013994617998407104

Linear Regression

# make linear regression data
from lib.data import makeRegressionData
data = makeRegressionData(100, degree=1)

X = data[:, :-1]
y = data[:, -1]

# initialize linear regression
from lib.regression import LinearRegression
w0, loss = LinearRegression(X, y)

# print the weights vector and the loss
print(w0, loss)

# plot the decision boundary
from lib.plots import plotRegressionLine2D
plotRegressionLine2D(X, y, w0)

Results

[-0.0305375   2.02728691] 0.030354517088163824

Polynomial Regression

# make polynomial regression data
from lib.data import makeRegressionData
data = makeRegressionData(100, degree=3)

X = data[:, :-1]
y = data[:, -1]

# transform the data to a higher dimension
from lib.transformation import polyMap
X3 = polyMap(X, 3)

# initialize polynomial regression
from lib.regression import LinearRegression
w0, loss = LinearRegression(X3, y)

# print the weights vector and the loss
print(w0, loss)

Results

[-0.00370442 -0.02731372 -0.00370442 -0.02731372 -0.02731372 -0.01275648
 -0.00370442 -0.02731372 -0.01275648  2.12742621] 0.029404600523786663