CS762 Fall 2021

Tentative Schedule (Subject to Change)

Week	Date	Topic	Reading materials	Assignments
1	Thursday, September 9	Course overview and introduction		Sign up for paper presentations and scribes (link to Google sheet)
2	Tuesday, September 14	Neural Architecture Design I (lecture)	Goodfellow-Bengio-Courville Chapter 6
2	Thursday, September 16	Neural Architecture Design II	ImageNet Classification with Deep Convolutional Neural Networks Deep Residual Learning for Image Recognition Densely Connected Convolutional Networks (to present) Mask R-CNN
3	Tuesday, September 21	Neural Architecture Design III	An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale (to present) Attention Is All You Need
3	Thursday, September 23	Trustworthy Deep Learning I (lecture)	Goodfellow-Bengio-Courville Chapter 7.5, 7.13	Team registration (link to Google sheet)
4	Tuesday, September 28	Trustworthy Deep Learning II	Out-of-distribution Reliability Deep Neural Networks are Easily Fooled: High Confidence Predictions for Unrecognizable Images Enhancing The Reliability of Out-of-distribution Image Detection in Neural Networks A Simple Unified Framework for Detecting Out-of-Distribution Samples and Adversarial Attacks Energy-based Out-of-distribution Detection (to present) MOS: Towards Scaling Out-of-distribution Detection for Large Semantic Space MOOD: Multi-level Out-of-distribution Detection
4	Thursday, September 30	Trustworthy Deep Learning III	Adversarial Robustness Intriguing properties of neural networks Towards Deep Learning Models Resistant to Adversarial Attacks Robust Physical-World Attacks on Deep Learning Models Unlabeled Data Improves Adversarial Robustness Defense Against Adversarial Images using Web-Scale Nearest-Neighbor Search Understanding and Mitigating the Tradeoff Between Robustness and Accuracy (to present)
5	Tuesday, October 5	No class (use the time to discuss the project proposal with team)		Project proposal submission deadline on October 8 Midnight (download proposal latex template here)
5	Thursday, October 7	Trustworthy Deep Learning IV	Fairness and Group Robustness Equality of Opportunity in Supervised Learning (to present) An investigation of why overparameterization exacerbates spurious correlations Distributionally Robust Neural Networks for Group Shifts: On the Importance of Regularization for Worst-Case Generalization Model Patching: Closing the Subgroup Performance Gap with Data Augmentation
6	Tuesday, October 12	Meetings with Instructor to Review Project Proposals (optional)		Signup sheet for 15-min slot (link to Google sheet)
8	Thursday, October 14	Interpretable Deep Learning I (Lecture)
9	Tuesday, October 19	Interpretable Deep Learning II	Learning Deep Features for Discriminative Localization (CAM) Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization (to present) Network Dissection: Quantifying Interpretability of Deep Visual Representations Interpretation of Neural Networks is Fragile
10	Thursday, October 21	Deep Learning Generalization and Theory I (Guest lecture: Tengyu Ma, Stanford University)
10	Tuesday, October 26	Deep Learning Generalization and Theory II	Exploring Generalization in Deep Learning Towards Understanding the Role of Over-Parametrization in Generalization of Neural Networks Visualizing the Loss Landscape of Neural Nets Reconciling modern machine-learning practice and the classical bias–variance trade-off Deep Double Descent: Where Bigger Models and More Data Hurt (to present)
11	Thursday, October28	Deep Learning Generalization and Theory III	Generalization and domain shift Invariant Risk Minimization (to present) Self-training Avoids Using Spurious Features Under Domain Shift
6	Tuesday, November 2	Learning with Less Supervision I (Lecture)
7	Thursday, November 4	Learning with Less Supervision II	Large-scale pre-training Exploring the Limits of Weakly Supervised Pretraining Big Transfer (BiT): General Visual Representation Learning Revisiting Unreasonable Effectiveness of Data in Deep Learning Era Rethinking Pre-training and Self-training Learning Transferable Visual Models From Natural Language Supervision (to present)
7	Tuesday, Novermber 9	Learning with Less Supervision III	Self-supervised learning Momentum Contrast for Unsupervised Visual Representation Learning A Simple Framework for Contrastive Learning of Visual Representations Supervised Contrastive Learning (to present) Big Self-Supervised Models are Strong Semi-Supervised Learners Understanding Contrastive Representation Learning through Alignment and Uniformity on the Hypersphere Understanding self-supervised Learning Dynamics without Contrastive Pairs
8	Thursday, November 11	Learning with Less Supervision IV	Data Augmentation Automating the Art of Data Augmentation: Overview Automating the Art of Data Augmentation: Practice Automating the Art of Data Augmentation: Theory Automating the Art of Data Augmentation: New Direction Unsupervised Data Augmentation for Consistency Training (to present)
11	Tuesday, November 16	Continual / Lifelong Learning I (Lecture)
12	Thursday, November 18	Continual / Lifelong Learning II	iCaRL - Incremental Classifier and Representation Learning LwF-Learning without Forgetting Overcoming catastrophic forgetting in neural networks SupSup - Supermasks in Superposition (to present)
12	Tuesday, November 23	Continual / Lifelong Learning III	Open-world recognition Towards Open World Object Detection(to present) Towards Open World Recognition
13	Thursday, November 25	No class (Thanksgiving)
13	Tuesday, November 30	Deep Generative Model I (Lecture)	Goodfellow-Bengio-Courville Chapter 20
14	Thursday, December 2	Deep Generative Model II	Generative Adversarial Networks InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets Stacked Generative Adversarial Networks Large Scale GAN Training for High Fidelity Natural Image Synthesis (to present) Wasserstein GAN Self-Attention Generative Adversarial Networks
14	Tuesday, December 7	Deep Generative Model III	Style transfer Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks StarGAN v2: Diverse Image Synthesis for Multiple Domains A Style-Based Generator Architecture for Generative Adversarial Networks Swapping Autoencoder for Deep Image Manipulation Frequency Domain Image Translation: More Photo-realistic, Better Identity-preserving (to present)
15	Thursday, December 9	Final project presentation (Part I)
15	Tuesday, December 14	Final project presentation (Part II)
16	Tuesday, December 21	Final project written report due (by end of the day)