Harvard Medical AI: Kathryn Wantlin on Unlabeled Data to Predict Out-of-Distribution performance

preview_player
Показать описание
A talk hosted by the Rajpurkar Lab at Harvard which works on developing medical AI. These talks cover recent papers or topics in core AI / medical AI in a format targeted to those interested in the cutting edge of AI and its applications in medicine.

In this session, lab member Kathryn Wantlin presents "Leveraging Unlabeled Data to Predict Out-of-Distribution Performance" by CMU and Google Brain.

Garg, S., Balakrishnan, S., Lipton, Z. C., Neyshabur, B., & Sedghi, H. (2022). Leveraging unlabeled data to predict out-of-distribution performance. arXiv preprint arXiv:2201.04234.

Subscribe to get weekly updates.
  1. Harvard Medical AI | Rajpurkar Lab
Рекомендации по теме
Harvard Medical AI: 0:28:53

Harvard Medical AI: Kathryn Wantlin on Identifying and Mitigating Model Failures

Harvard Medical AI: 0:20:54

Harvard Medical AI: Kathryn Wantlin on Unlabeled Data to Predict Out-of-Distribution performance

Harvard Medical AI: 0:34:17

Harvard Medical AI: Kathryn Wantlin on 'Emerging Properties in Self-Supervised Vision Transform...

Self-Supervised Correspondence Estimation 0:04:01

Self-Supervised Correspondence Estimation via Multiview Registration (WACV 2023)

Explaining Deep Convolutional 0:05:01

Explaining Deep Convolutional Neural Networks via Latent Visual Semantic Filter | CVPR 2022

Learning from Unlabeled 1:23:19

Learning from Unlabeled Videos for Recognition, Prediction, and Control

[CVPR'23] Distilling Self-Supervised 0:08:00

[CVPR'23] Distilling Self-Supervised ViTs for Weakly-Supervised Few-Shot Classification Segment...

What if there's 0:19:26

What if there's a better way to monitor models than measuring distribution drift?

CAP6412 2022: Lecture 0:26:24

CAP6412 2022: Lecture 25 - Emerging Properties in Self-Supervised Vision Transformers

CVPR 2023 - 0:08:00

CVPR 2023 - CrOC: Cross-View Online Clustering for Dense Visual Representation Learning

2015 WVYS Chamber 0:09:05

2015 WVYS Chamber Concert: Quartet Tres Fabuleux

DINO: Emerging Properties 0:08:14

DINO: Emerging Properties in Self-Supervised Vision Transformers (paper illustrated)

Rethinking Self-supervised Correspondence 0:10:33

Rethinking Self-supervised Correspondence Learning: A Video Frame-level Similarity Perspective

LEAD: Self-Supervised Landmark 0:04:47

LEAD: Self-Supervised Landmark Estimation by Aligning Distributions of Feature Similarity