Neural Network-Based Function Approximation

The project implements a Multi-Layer Perceptron (MLP) to approximate complex mathematical functions using deep learning. The model is designed to learn and predict a nonlinear function with noise, leveraging various optimization algorithms such as SGD, Momentum, RMSprop, and Adam.

Features:
- Custom MLP Implementation: developed a neural network from scratch using NumPy
- Multiple Optimization Techniques: implemented different optimization algorithms to enhance training efficiency
- Loss & R² Tracking: monitored performance during training using Mean Squared Error (MSE) and R² score
- Visualization & Analysis: generated 3D plots and learning curves to evaluate model performance
- Model Persistence: enabled saving and loading of trained models for reproducibility

This project demonstrates my ability to build, train, and analyze deep learning models without relying on high-level libraries like TensorFlow or PyTorch, showcasing a strong understanding of neural networks at a fundamental level.