Machine Learning From Scratch
- Non-Linear Regression
- Polynomial regression for a projectile motion problem with batch, mini-batch, and stochastic gradient descent, including cost function plots and variable learning rates.
- Multivariate Regression
- Predicting house prices using gradient descent with multiple features.
- Binary Classification
- Rainfall prediction with logistic regression, binary cross-entropy loss, and scipy optimization.
- Multi-Class Classification
- MNIST Dataset classification using binary cross-entropy loss and one-hot encoding.
- Softmax Multi-Class Classification
- MNIST Fashion dataset classification with batch gradient descent and accuracy comparison.
- Neural Network from Scratch
- Numpy-based multi-layer perceptron for deep learning insights.
- Convolution Neural Network from Scratch
- CNN built from scratch with convolution, max-pooling, softmax, and gradients.
- Recurrent Neural Network from Scratch
- RNN implementation in NumPy with Tanh activation.