MADLab 2022 Summer Workshop
Dates and Location:
June 14-15, 2022
6/14 @ the John Crerar Library (home of UChicago CS Department), 5730 S Ellis Ave, Chicago, IL 60637
6/15 @ the Toyota Technological Institute at Chicago, 6045 S Kenwood Ave, Chicago, IL 60637
Hotels:
- The Study At Univ Of Chicago — right on campus, halfway between CS department and TTIC. For reserving at the military rate at The Study, guests can contact their reservations department directly at reservations.chicago@studyhotels.com or by calling 773-643-1600. The rate will be $216 for June.
- Sophy Hyde Park — off campus but very close; close to restaurants, bars, shops
- Hyatt Place — off campus but very close; close to restaurants, bars, shops. UChicago Friend and Family discount available; rate would be $178. To book online using the government rate at the Hyatt, guests simply select “Government” under ‘Special Rates’ to view/book this; for June that rate is $212.
Banquet:
Planned for Promontory Point, right on Lake Michigan.
Schedule:
Day 1, June 14th (@ John Crerar Library)
8:45 – 9:15 Coffee, gathering
9:15 – 9:30 Welcoming remarks (Becca, Rob, Lee, Erik)
9:30 – 10:00 Talk 1 on Data-Efficient ML (Jerry Zhu): Efficient Data Collection for Learning Requires Incentives
10:00 – 10:30 Talk 2 on Data-Efficient ML (“Interactive Learning for Mission Planning” by Scott Sievert)
10:30 – 10:45 Break
10:45 – 12:00 Day 1 Lightning talks
12:00 – 13:30 Lunch
13:30 – 14:00 Talk 3 on Operational Robustness (Karen)
14:00 – 14:30 Talk 4 on Operational Robustness (Ashley Prater-Bennette – video)
14:30 – 15:00 Break
15:00 – 17:00 Day 1 Poster session
18:00 – 21:00 Banquet
Day 1 Lightning Talks
- Yuxin Chen, UChicago, Teaching an Active Learner with Contrastive Examples
- Rob Nowak, U-W Madison, Training Out-of-Distribution Detectors in the Wild
- Rebecca Willett, UChicago, Auto-differentiable Ensemble Kalman Filters
- Andre Beckus, AFRL, A methodology for flattening the command and control problem space
- Yinglun Zhu, UW-Madison, Towards Statistical and Computational Efficiencies in Active Learning
- Greg Canal, UW-Madison, One for All: Simultaneous Metric and Preference Learning over Multiple Users
- Simon Khan, AFRL, User Authentication by Fusion of Mouse Dynamics and Widget Interactions: Two Experiments with PayPal and Facebook
- Jerry Zhu, UW-Madison, Creative Bandit: Generating More and Better Training Data
- Jake Soloff, UChicago, TBD
- Erik Blasch, AFOSR, Motivation towards Certifiable AI
- Zhenmei Shi, UW-Madison, A Theoretical Analysis on Feature Learning in Neural Networks: Emergence from Inputs and Advantage over Fixed Features; Towards Evaluating the Robustness of Neural Networks Learned by Transduction
Day 1 Posters
- Daniel Park, AFRL, Towards Cost-Constrained Adversarial Examples
- Claire Hardy, AFRL, Common and Uncommon Feature Extractor Robustness Against Novel Input
- Walter Bennette, AFRL, Hierarchical Open-Set Recognition
- Sean Sisti, AFRL, TBD
- Amy Wong, AFRL, KnowML
- Young Wu, UW-Madison, MARL Poisoning
- Elena Orlova, UChicago, ML for Subseasonal forecasting
- Nathan McDonald, AFRL, Online Learning with Hyperdimensional …
- Ruoxi Jiang, UChicago, Embed and Emulate: Learning to estimate parameters of dynamical systems with uncertainty quantification
- Jayaram Raghuram, UW-Madison, Stratified Adversarial Robustness with Rejection
- Cody Kearse, AFRL, Modularized Class Specific Neural Network Configurations for Precision Training
- Alex Hildenbrandt & Ashley Diehl, AFRL, Robust and Secure Machine Learning
- Jeremy McMahan, UW-Madison, Offline Data Poisoning in MARL
- Ankita Pasad, TTIC, Layer-Wise Analysis Of Self-Supervised Speech Models
- Danica Fliss, UW-Madison, Plug-and-Play Methods for Satellite Image Restoration
Day 2, June 15th (@ TTIC):
9:00 – 9:30 Coffee, gathering
9:30 – 10:00 Talk 5 on Adversarial Robustness (Yingyu)
10:00 – 10:30 Talk 6 on Adversarial Robustness (Ryan Luley)
10:30 – 10:45 Break
10:45 – 12:00 Day 2 Lightning talks
12:00 – 13:30 Lunch + Day 2 Poster session
13:30 – 14:00 Talk 7 on Computationally Efficient ML (Lipasti): PrGEMM: Hardware Acceleration for Sparse Matrix Multiply
14:00 – 14:30 Talk 8 on Computationally Efficient ML (Bai): Toward Intelligence in Communication Networks.
14:30 – 15:15 Panel
15:15+ Unstructured research meetings
Day 2 Lightning Talks
- Jeremy McMahan, UW-Madison, Offline Data Poisoning in MARL
- Benjamin Robinson, AFRL, High-Dimensional Principal Components and Signal Detection
- Jy-yong Sohn, UW-Madison, GenLabel: Mixup Relabeling using Generative Models
- David Uminsky, UChicago, TBD
- Shageenth Sandrakumar, AFRL, TBD
- Kangwook Lee, UW-Madison, TBD
- Michael Maire, UChicago, Faster Training by Growing Deep Networks
- Alex Alaniz, AFRL, Weapons engagement OPtimizeR (WOPR)
- Cong Ma, UChicago, Scaling and Scalability: Provable Nonconvex Low-Rank Tensor Estimation from Incomplete Measurements
- Shubham Kumar Bharti, UW-Madison, Towards provable defenses against trojan attacks in RL
Day 2 Posters
- Simon Khan, AFRL, User Authentication by Fusion of Mouse Dynamics and Widget Interactions: Two Experiments with PayPal and Facebook
- Zhenmei Shi, UW-Madison, A Theoretical Analysis on Feature Learning in Neural Networks: Emergence from Inputs and Advantage over Fixed Features; Towards Evaluating the Robustness of Neural Networks Learned by Transduction
- Clare Thiem, AFRL, NeuroPipe: A combined Development Pipeline for Novel Neuromorphic Hardware
- Andre Beckus, AFRL, Methodology for Flattening the Command and Control Problem Space
- Kartik Sreenivasan, UW-Madison, Rare Gems: Finding Lottery Tickets at Initialization
- Kang Jun Bai, AFRL, Toward Intelligence in Communication Networks
- Cong Ma, UChicago, Scaling and Scalability: Provable Nonconvex Low-Rank Tensor Estimation from Incomplete Measurements
- Shubham Kumar Bharti, UW-Madison, Towards provable defenses against trojan attacks in RL
- Chien-Fu Chen, UW-Madison, Poster: PrGEMM: A Parallel Reduction SpGEMM Accelerator
- Xiaodan Du, TTIC, Learning a Natural Language Interface for Pretrained Image Generators
- Jiahao Li, TTIC, Adapting CLIP For Phrase Localization Without Further Training
- David Yunis, TTIC, Singular dynamics and linear mode connectivity in neural networks
- Junyi Wei, UW Madison, An efficient self-supervised training method for Detr