Local Reference Geometry Residual Augmentation for Imbalanced Time Series Classification

Name: Local Reference Geometry Residual Augmentation for Imbalanced Time Series Classification
Uploaded: 2026-09-06T22:47:34.987589+00:00
Duration: 472 s
Description: The episode discusses 'Local Reference Geometry Residual Augmentation for Imbalanced Time Series Classification.' Hosts explain how this method addresses local geometry failures in imbalanced data, where minority samples are poorly supported by neighbors. The solution, LRG, is a post-hoc augmentation that improves model robustness without retraining the original encoder.

The episode discusses 'Local Reference Geometry Residual Augmentation for Imbalanced Time Series Classification.' Hosts explain how this method addresses local geometry failures in imbalanced data, where minority samples are poorly supported by neighbors. The solution, LRG, is a post-hoc augmentation that improves model robustness without retraining the original encoder.

Open the full episode