MLAlgorithms

Implementation of machine learning algorithms from scratch for learning purposes. PyTorch is the only dependency. Scikit-learn and Matplotlib are used in the example.py file to compare the scikit-learn implementations with the custom made implementations and give examples on how to use the components. However, the library itself only depends on PyTorch.

Requirements

• torch (2.4.0)
• [Optional] : scikit-learn (1.5.1)

Installation

Clone the project and use algorithms. The example.py file shows how to use the algorithms.

pip install torch==2.4.0

git clone git@github.com:GM07/MLAlgorithms.git
cd MLAlgorithms

Algorithms implemented

Regression

• Linear Regression
• Ridge Regression
• Lasso Regression
• Logistic Regression

Clustering

• KMeans

• DBSCAN

Naive Bayes

• Gaussian Naive Bayes
• Bernoulli Naive Bayes
• Multinomial Naive Bayes

Dimensionality Reduction

• PCA
• t-SNE
• UMAP

Decision Trees

• Decision Tree Classifier
• Decision Tree Regressor

Ensemble Learning

• Bagging
• Boosting
• XGBoost
• Random Forests

Deep Learning Layers

• Linear Layer
• Attention Layer
• Convolutional Layer
• Max Pooling Layer
• Average Pooling Layer
• Layer Normalization Layer
• Batch Normalization Layer

Deep Learning

• Neural Network
• LSTM
• RNN
• Transformer
• Diffusion Model

Metrics

• RMSE

References

• Scikit-learn: Machine Learning in Python, Pedregosa et al., JMLR 12, pp. 2825-2830, 2011.