Implementing Machine Learning and Deep Learning algorithms completely from scratch using only Python, NumPy, and Matplotlib.
The purpose of this repository is to understand the mathematics and implementation details behind Machine Learning algorithms before relying on high-level frameworks such as Scikit-Learn, TensorFlow, or PyTorch.
- Pure Python implementation
- NumPy-based numerical computation
- No Machine Learning libraries
- Mathematical implementation
- Gradient Descent optimization
- Visualization using Matplotlib
- Educational code with clean structure
| Algorithm | Description | Status |
|---|---|---|
| Linear Regression | Predict continuous values using Gradient Descent | ✅ |
| Logistic Regression | Binary Classification using Sigmoid Function | ✅ |
| Algorithm | Status |
|---|---|
| K-Nearest Neighbors (KNN) | ⏳ |
| Decision Tree | ⏳ |
| Random Forest | ⏳ |
| Naive Bayes | ⏳ |
| Support Vector Machine (SVM) | ⏳ |
| Perceptron | ⏳ |
| Neural Networks | ⏳ |
| Convolutional Neural Networks (CNN) | ⏳ |
| Recurrent Neural Networks (RNN) | ⏳ |
| Attention Mechanism | ⏳ |
| Transformer | ⏳ |
Linear Regression predicts continuous numerical values by fitting a straight line to the training data.
Mean Squared Error (MSE)
Gradient Descent
Logistic Regression is a binary classification algorithm that predicts probabilities using the Sigmoid activation function.
Sigmoid
Binary Cross Entropy (BCE)
Gradient Descent
This repository is part of my journey to deeply understand Artificial Intelligence by implementing every Machine Learning and Deep Learning algorithm from scratch.
Instead of using high-level Machine Learning libraries, every model is built step by step using only Python and NumPy to better understand the underlying mathematics and optimization process.
The repository will continue to grow as more algorithms are implemented.