Course Demonstrations

EE 541: Deep Learning Foundations

Notebooks and code demos. You may copy and modify for use in EE 541.

Week 1: Introduction to Deep Learning

• Minimal PyTorch Example - Two-layer network on synthetic data, SGD optimization
• Fashion-MNIST Classification - CNN architecture, training loop, accuracy metrics
• TensorBoard Visualization - Loss curves, weight histograms, computational graph

Week 3: MMSE Estimation

• Scalar MMSE Examples - Uniform, exponential, and Gaussian cases with analytical solutions
• MMSE Estimator Implementation - NumPy implementation for arbitrary joint distributions
• Gaussian Equivalence - Proof that MMSE equals LMMSE for jointly Gaussian variables
• Linear vs Nonlinear MMSE - MSE comparison on sinusoidal and polynomial relationships
• LMS Adaptive Filtering - Step-size analysis, convergence behavior, tracking performance

Week 9: Training Deep Neural Networks

• Minimal MLP Example - Two-layer network on FashionMNIST with training loop

Week 10: Convolutional Neural Networks

• FashionMNIST CNN - Mini-VGG with batch normalization and dropout
• Dogs vs Cats CNN - Binary classification on RGB images with data augmentation