Deep Learning from Scratch - Lab-based Learning

Hands-on Deep Learning course with Python + Numpy, from basic concepts to practical implementation.

🎯 Objectives

• Understand deeply how Neural Networks work
• Code from scratch without frameworks (TensorFlow, PyTorch)
• Master the mathematics behind Deep Learning
• Learn through systematic hands-on labs

📚 Lab Structure

Lab 01: Fundamental Concepts 🔥

• Forward Propagation
• Loss Functions (MSE)
• Backward Propagation
• Gradient Descent

---

Lab 02: Activation Functions

• Sigmoid, ReLU, Tanh
• Non-linearity in Neural Networks
• Vanishing gradient problem
• See: lab02/

Lab 03: Multi-layer Networks

• Stacking layers
• Deep neural networks
• Backpropagation through multiple layers
• See: lab03/

Lab 04: Classification Problems

• Binary classification
• Softmax and Cross-entropy
• Multi-class classification
• See: lab04/

Lab 05: Optimization Algorithms

• SGD variations
• Momentum, Adam, RMSprop
• Learning rate scheduling
• See: lab05/

Lab 06: Regularization

• Overfitting problem
• L1/L2 regularization
• Dropout
• See: lab06/

Lab 07: Real Datasets

• MNIST handwritten digits
• Data preprocessing and normalization
• Model evaluation metrics
• See: lab07/

---

🚀 Getting Started

cd lab01/
# Start with the first lab