Module 3
Understand neural networks, how they learn, and build your first deep learning models with modern frameworks.
Understand how multi-layer neural networks learn hierarchical representations. Master backpropagation, loss landscapes, and PyTorch implementations.
Understand the limits of manual feature engineering and why deep hierarchical neural networks revolutionized artificial intelligence.
Explore the biological inspiration and mathematical formulation of the Perceptron, the atomic building block of deep learning.
Master the mathematical non-linearities that allow neural networks to bend space: Sigmoid, Tanh, ReLU, Leaky ReLU, and Softmax.
Understand the anatomy of Multi-Layer Perceptrons (MLP): input layers, hidden layers, output layers, and parameter dimensions.
Trace data through the computational graph from input tensors to output activations via vectorized matrix multiplications.
Understand how mathematical loss functions measure prediction error and guide neural network optimization.
Unravel the mathematical core of deep learning: computing analytical gradients via the multivariable Chain Rule and stepping down loss landscapes.
Explore Google’s enterprise deep learning framework: static computational graphs, high-level Keras Sequential API, and TensorBoard.
Master PyTorch, the favorite framework of AI researchers worldwide: dynamic computation graphs (Eager Mode), Tensors, and nn.Module.
Synthesize Module 3: construct, train, validate, and evaluate a multi-layer deep neural network on handwritten digits (MNIST).