CPTS 434/534 – Neural Network Design and Application

Course Overview

CPTS 434/534 introduces the foundations and practical aspects of deep neural networks, a cornerstone of modern machine learning. The course blends theoretical concepts in statistical learning with hands-on implementation of key deep learning models.

Learning outcomes:

  • Principles of representation learning and optimization
  • Neural network architectures (MLPs, CNNs, RNNs, transformers)
  • Practical training techniques and modern model variants

By the end of the course, students will understand how deep models are designed, trained, and applied to real-world data across vision, sequence, and structured domains.

Syllabus

Lecture Schedule

Lecture Topic Slides
1 [0] Syllabus and Overview  
2 [ML] ML Basics Slides
3 [ML] PAC Learning Slides
4 [ML] ERM and MLE Slides
5 [NN] NN Basics Slides
6 [NN] Compositing Units in NN Slides
7 [NN] MLP Softmax Classifier Slides
8 [CNN] Convolutional Layer and CNN Slides
9 [CNN] CNN Architectures Slides
10 [CNN] Practical Training of CNN Slides
11 [RNN] Sequence Data and RNN Slides
12 [RNN] LSTM and RNN Architecutres Slides
13 [ModernNN] Attention Mechanism  
14 [ModernNN] Transformer and ViT  
15 [GNN] Graph Data and Representation  
16 [GNN] GNN  
17 [GenAI] GAN and VAE  
18 [GenAI] Diffusion Models  
19 [FM] Foundation Models  
20 [FM] Multimodal Foundation Models  
21 [FL] Federated Learning  
22 [UQ] Conformal Prediction and Mean-Risk Model  
23 [Agent] Agentic Workflow