paper

Range Penalization: Theoretical Insights with Applications in Federated Learning

paperactiveprovisionalrange-penalization-theoretical-insights-with-applications-in-federated-learning-1e10800f·1 events·first seen 7d ago

Aliases: Range Penalization: Theoretical Insights with Applications in Federated Learning

More like this (12)

Federated Learning Heterogeneous Differential Privacy Federated Learning Your Privacy My Cloak: Backdoor Attacks on Differentially Private Federated Learning FlashbackCL: Mitigating Temporal Forgetting in Federated Learning TailLoR: Protecting Principal Components in Parameter-Efficient Continual Learning FFR: Forward-Forward Learning for Regression Entropy-Regularized Reinforcement Learning PC Layer: Polynomial Weight Preconditioning for Improving LLM Pre-Training Rank-Constrained Subspace Learning (RCSL)Language Model Finetuning ExpRL: Exploratory RL for LLM Mid-Training CLP: Collocation-Length Prediction for Zero-Loss Adaptive Multi-Token Inference

Recent events (1)

3arXiv · cs.LG·7d ago·source ↗

Range Penalization for Federated Learning: Polar Clustering and Statistical Accuracy

This paper introduces range regularization for federated learning, identifying shared-weight features across clients while adaptively clustering personalized feature weights at extreme values (termed polar clustering). The approach targets statistical accuracy, cross-client regularity, and resource efficiency for quantization and coding. New nonasymptotic proof techniques are developed for the seminorm-based regularizer, alongside a fast optimization algorithm exploiting local strong convexity.

Inference Economics Range Penalization: Theoretical Insights with Applications in Federated Learning