Washington University in St. Louis

Instructor: Jeff Heaton

• Section 1. Summer 2026, Monday, 6:00 PM
  Location: online

Course Description

Deep learning is a group of exciting new technologies for neural networks. Through a combination of advanced training techniques and neural network architectural components, it is now possible to create neural networks that can handle tabular data, images, text, and audio as both input and output. Deep learning enables a neural network to learn hierarchies of information in a manner similar to the way the human brain functions. This course will introduce the student to classic neural network structures, Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTMs), Gated Recurrent Units (GRUs), Generative Adversarial Networks (GANs), and reinforcement learning. Application of these architectures to computer vision, time series, security, natural language processing (NLP), and data generation will be covered. High-Performance Computing (HPC) aspects will demonstrate how deep learning can be leveraged on both graphical processing units (GPUs) and grids. The focus is primarily on the application of deep learning to problems, with some introduction to the mathematical foundations. Students will use Python and PyTorch to implement deep learning. It is not necessary to know Python prior to this course; however, familiarity with at least one programming language is assumed. This course will be delivered in a hybrid format, combining classroom and online instruction.

Objectives

1. Explain how neural networks (deep and otherwise) compare to other machine learning models.
2. Determine when a deep neural network would be a good choice for a particular problem.
3. Demonstrate understanding of the material through applied programming assignments and a Kaggle competition.

Syllabus

This syllabus presents the expected class schedule, due dates, and reading assignments.
Download current syllabus

Module	Content
Module 1 Meet on 08/24/2026	Module 1: Introduction to Neural Networks 1.1 Course Overview 1.2 AI and Neural Networks 1.3 Modern Neural Networks 1.4 Deep Learning in the Age of GenAI 1.5 Vibe Coding: Acceptable AI Use in this Course We will meet on campus this week (in-class meeting #1)
Module 2 Week of 08/31/2026	Module 2: PyTorch for Neural Networks 2.1 Numeric Processing with PyTorch 2.2 Deep Learning with PyTorch 2.3 Preprocessing for PyTorch 2.4 nn.Module vs nn.Sequential for PyTorch 2.5 Beyond the CPU Module 1 Program due: 09/01/2026 Icebreaker due: 09/01/2026
Module 3 Week of 09/14/2026	Module 3: Working with Tabular Data 3.1 Should Neural Networks be Used for Tabular Data 3.2 Pandas for PyTorch 3.3 Encoding Categorical Values in Pandas 3.4 Grouping, Sorting and Shuffling 3.5 Feature Engineering Module 3 Program due: 09/15/2026
Module 4 Meet on 09/21/2026	Module 4: Training 4.1 An Overview of PyTorch Persistence 4.2 Early Stopping and Network Checkpointing 4.3 Using K-Fold Cross-validation with PyTorch 4.4 Training Schedules 4.5 Regularization and Dropout Module 4 Program due: 09/22/2026 We will meet on campus this week (in-class meeting #2)
Module 5 Week of 09/28/2026	Module 5: Convolutional Neural Networks 5.1 PyTorch Image Processing 5.2 Convolutional Neural Networks (CNNs) 5.3 Augmentation 5.4 Pretrained CNN Models 5.5 YOLO Module 5 Program due: 09/29/2026
Module 6 Meet on 10/05/2026	Module 6: Time Series Data 6.1 Time Series Introduction 6.2 LSTM for Time Series 6.3 Transformers for Time Series 6.4 Prophet 6.5 Toy SP500 Predictor Module 6 Program due: 10/06/2026 We will meet on campus this week
Module 7 Week of 10/12/2026	Module 7: PyTorch Building Blocks 7.1 Model Structure 7.2 Learnable Layers 7.3 Nonlinearities (Activations) 7.4 Normalization & Regularization 7.5 Shapes Module 7 Program due: 10/13/2026
Module 8 Meet on 10/19/2026	Module 8: Kaggle and Hyperparameter Tuning 8.1 Introduction to Kaggle 8.2 Ensemble Learning 8.3 Hyperparameter Tuning 8.4 Sample Kaggle Solution 8.5 Current Semester's Kaggle Module 8 Program due: 10/20/2026 We will meet on campus this week
Module 9 Week of 10/26/2026	Module 9: Foundations of Generative AI 9.1 Transformer Architecture and Its Modern Variants 9.2 Pretrained Models and the Hugging Face Ecosystem 9.3 Embeddings and Vector Representations 9.4 Diffusion Models and Image Generation 9.5 Accessing API-Based LLMs Module 9 Program due: 10/27/2026
Module 10 Week of 11/02/2026	Module 10: Facial Recognition 10.1 Detecting Faces in an Image 10.2 Detecting Facial Features 10.3 Image Augmentation 10.4 Emotion and Sentiment Detection 10.5 Introduction to Dynaface Module 10 Program due: 11/03/2026
Module 11 Week of 11/09/2026	Module 11: Troubleshooting and Evaluating PyTorch Models 11.1 Debugging with LLMs 11.2 Critical PyTorch Errors 11.3 Overfitting and Underfitting 11.4 Vanishing and Exploding Gradients 11.5 Error Metrics Beyond Accuracy Module 11 Program due: 11/10/2026
Module 12 Week of 11/16/2026	Module 12: Explainability 12.1 What is Explainability 12.2 Feature Importance 12.3 Visualization for Computer Vision 12.4 Gradient-Based Explanations 12.5 Limitations of Explainability
Module 13 Week of 11/23/2026	Module 13: Other Techniques 13.1 Autoencoders and Dimensionality Reduction 13.2 Anomaly Detection 13.3 Reinforcement Learning with TorchRL 13.4 Pruning Neural Networks 13.5 Self-Supervised Learning Kaggle Competition Closes: 11/29/2026 (midnight) Kaggle Assignment due in Canvas: 12/01/2026
Module 14 Week of 11/30/2026	Module 14: Wrapping Up 14.1 Model Drift 14.2 Dealing with Bias 14.3 Other Deep Learning Frameworks 14.4 Deploying a PyTorch Neural Network 14.5 Future of AI

Datasets

• Datasets can be downloaded here