Schedule
May 5, 2023
May 5, 2023
9:00 - 9:10
Introduction and opening remarks
Introduction and opening remarks
9:10 - 9:45
Invited talk 1 + Q&A
Invited talk 1 + Q&A
Praneeth Netrapalli (Towards neural networks robust to distribution shifts)
9:45 - 10:20
Invited talk 2 + Q&A
Invited talk 2 + Q&A
Vitaly Feldman (What Neural Networks Memorize and Why)
10:35 - 10:45
Contributed talk 2
Contributed talk 2
On the Efficacy of Differentially Private Few-shot Image Classification
On the Efficacy of Differentially Private Few-shot Image Classification
10:45 - 10:55
Contributed talks 3
Contributed talks 3
Practical Differentially Private Hyperparameter Tuning with Subsampling
Practical Differentially Private Hyperparameter Tuning with Subsampling
10:55 - 11:05
Contributed talk 4
Contributed talk 4
Error Discovery by Clustering Influence Embeddings
Error Discovery by Clustering Influence Embeddings
11:05 - 11:15
Coffee break
Coffee break
12:15 - 13:40
Lunch break
Lunch break
13:40 - 14:15
Invited talk 3 + Q&A
Invited talk 3 + Q&A
Fereshte Khani (Impacts of Data Scarcity on Groups and Harnessing LLMs for Solution)
14:15 - 14:50
Invited talk 4 + Q&A
Invited talk 4 + Q&A
Ruth Urner (How (not) to Model an Adversary)
14:50 - 15:25
Invited talk 5 + Q&A
Invited talk 5 + Q&A
Nicholas Carlini (Practical poisoning of machine learning models)
15:25 - 15:35
Coffee break
Coffee break
15:35 - 16:05
Panel Discussion
Panel Discussion
16:05 - 16:15
Contributed Talk 5
Contributed Talk 5
Project with Source, Probe with Target: Extracting Useful Features for Adaptation to Distribution Shifts
Project with Source, Probe with Target: Extracting Useful Features for Adaptation to Distribution Shifts
16:15 - 16:25
Contributed Talk 6
Contributed Talk 6
Efficient utilization of pre-trained model for learning with noisy labels
Efficient utilization of pre-trained model for learning with noisy labels
16:25 - 16:30
Closing Remarks
Closing Remarks