T81 558: Applications of Deep Neural Networks

Washington University in St. Louis

Instructor: Jeff Heaton

• Section 1. Fall 2026, Tuesday, 2:30 PM
  Location: EADS, Room 00102 (hybrid)

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 08/25/2026	Module 1: Introduction to Neural Networks 1.1 Overview 1.2 AI and Neural Networks 1.3 Deep Learning and Neural Network Introduction 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 Meet on 09/01/2026	Module 2: PyTorch for Neural Networks 2.1 Numeric Processing with PyTorch 2.2 Deep Learning with PyTorch 2.3 Encoding a Feature Vector for PyTorch Deep Learning 2.4 Sequences vs Classes in PyTorch 2.5 Beyond the CPU Module 1 Program due: 09/02/2026 Icebreaker due: 09/02/2026 We will meet on campus this week (in-class meeting #2)
Module 3 Week of 09/08/2026	Module 3: Working with Tabular Data 3.1 Should Neural Networks be Used for Tabular Data 3.2 Categorical and Continuous Values 3.3 Grouping, Sorting, and Shuffling 3.4 Apply and Map 3.5 Feature Engineering Module 2 Program due: 09/09/2026
Module 4 Meet on 09/15/2026	Module 4: PyTorch Training 4.1 PyTorch Persistence 4.2 Early Stopping 4.3 K-Fold Cross-validation with PyTorch 4.4 Training Schedules 4.5 Regularization and Dropout Module 3 Program due: 09/16/2026 We will meet on campus this week (in-class meeting #3)
Module 5 Week of 09/22/2026	Module 5: Convolutional Neural Networks (CNN) for Computer Vision 5.1 Image Processing in Python 5.2 Convolutional Neural Networks for Images 5.3 Inside Augmentation 5.4 Transfer Learning for Computer Vision 5.5 Recognizing Multiple Images with YOLO Module 4 Program due: 09/23/2026
Module 6 Week of 09/29/2026	Module 6: Time Series in PyTorch 6.1 Time Series Data Encoding for Deep Learning, PyTorch 6.2 LSTM-Based Time Series with PyTorch 6.3 Transformers for Time Series in PyTorch 6.4 Predicting with Meta Prophet 6.5 Vibe Coding a Toy SP500 Model Module 5 Program due: 09/30/2026
Module 7 Week of 10/13/2026	Module 7: PyTorch Building Blocks 7.1 Model Structure 7.2 Learnable Layers 7.3 Nonlinearities (Activations) 7.4 Normalization & Regularization 7.5 Shape Module 6 Program due: 10/14/2026
Module 8 Meet on 10/20/2026	Module 8: Kaggle Data Sets 8.1 Introduction to Kaggle 8.2 Building Ensembles with Scikit-Learn and PyTorch 8.3 Architecting Network: Hyperparameters 8.4 Bayesian Hyperparameter Optimization for PyTorch 8.5 Current Semester's Kaggle Module 7 Program due: 10/21/2026 We will meet on campus this week (in-class meeting #4)
Module 9 Week of 10/27/2026	Module 9: Foundations of Generative AI 9.1 Introduction to Transformers 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 8 Program due: 10/28/2026
Module 10 Week of 11/03/2026	Module 10: Facial Processing 10.1 Detecting Faces in an Image 10.2 Detecting Facial Features 10.3 Reality Augmentation 10.4 Application: Emotion Detection 10.5 Dynaface Module 9 Program due: 11/04/2026
Module 11 Week of 11/10/2026	Module 11: Troubleshooting and Evaluating PyTorch Models 11.1 Debugging with PyTorch Networks 11.2 Critical PyTorch Errors 11.3 Overfitting and Underfitting 11.4 Vanishing and Exploding Gradients 11.5 Error Metrics Beyond Accuracy Module 10 Program due: 11/11/2026
Module 12 Week of 11/17/2026	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 11/24/2026	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 due: 12/04/2026
Week 14 Week of 12/01/2026	Wrapup Discuss final Kaggle results and future directions of this technology.

Datasets

• Datasets can be downloaded here