Nonparametric Distribution Matching for Self-Supervised Whole-Slide Image Condensation

Name: Nonparametric Distribution Matching for Self-Supervised Whole-Slide Image Condensation
Uploaded: 2026-10-19T08:47:26.409447+00:00
Duration: 1156 s
Description: The authors introduced NICER, a new method for condensing whole-slide images (WSIs) by framing it as a distribution-matching problem. Instead of simple heuristic methods, NICER uses a nonparametric prior to learn slide-adaptive feature representations. Experiments show NICER consistently outperforms prior techniques, improving accuracy by 7.44% while offering better efficiency.

The authors introduced NICER, a new method for condensing whole-slide images (WSIs) by framing it as a distribution-matching problem. Instead of simple heuristic methods, NICER uses a nonparametric prior to learn slide-adaptive feature representations. Experiments show NICER consistently outperforms prior techniques, improving accuracy by 7.44% while offering better efficiency.

Open the full episode