technique

Class-Balanced Reservoir Sampling

techniqueactiveprovisionalclass-balanced-reservoir-sampling-3d0cb8cc·1 events·first seen 14d ago

Aliases: Class-Balanced Reservoir Sampling

Co-occurring entities

CIFAR-100 CIFAR-10 FlashbackCL FlashbackCL: Mitigating Temporal Forgetting in Federated Learning Flashback

More like this (12)

Min-p sampling Best-of-N Sampling Top-p sampling Global-batch Load Balancing finite-sample posterior sampling framework gated demographic residual model diffusion posterior sampling EB-Sampler Proportional Allocation Rule importance sampling Log Probability Bias Analysis BERTScore

Recent events (1)

4arXiv · cs.LG·14d ago·source ↗

FlashbackCL extends federated learning to mitigate temporal distribution shift and forgetting

FlashbackCL is a proposed extension to the Flashback federated learning method that addresses temporal forgetting — the degradation caused by client data distributions drifting over time, a scenario existing FL methods do not handle. The approach introduces temporally-decayed label counts, a device-aware replay buffer with Class-Balanced Reservoir Sampling, and server-side coreset curation. On CIFAR-10 with 50 clients, FlashbackCL achieves 6.9–10.0% relative improvement over Flashback while reducing temporal forgetting by up to 68%, with CBRS replay identified as the critical component.

Evaluation and Benchmarking CIFAR-100 CIFAR-10 FlashbackCL +3 more