cs.LGNov 18, 2026

Kernel Ridge Attention: Nonlinear Memory with Selective Forgetting

Authors: Haowei Chen, Yichen Wang, Congfu Xu

Organizations: Independent Researcher · Cornell University · Zhejiang University

Abstract

Kernel Ridge Attention (KRA) is a recurrent sequence mixer that combines learned nonlinear features with covariance-guided memory updates. It stores associations through recursive ridge regression, using correlations among past keys to determine how new observations modify memory. A content-dependent Ornstein--Uhlenbeck transition controls retention and restores uncertainty for future writes. This construction admits an exact temporal-kernel interpretation and maintains an inference state whose size is independent of context length. Associative block maps and explicit adjoints support parallel training while preserving the recursive update. In controlled language-modeling experiments, KRA improves prediction quality over attention and recurrent baselines, strengthens natural associative recall over the evaluated recurrent models, and achieves practical distributed training throughput. These properties provide a principled foundation for nonlinear, adaptive sequence memory.

Figures & tables

Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. Kernelized Linear Attention: Breaking the Capacity Wall with Symmetric Cones

    Jul 19, 2026Ayoub Ghriss, Sourav ChakrabortyEfficient AttentionLinear Attention

  2. Hybrid Associative Memories

    Mar 20, 2026Leon Lufkin, Tomás Figliolia, Beren Millidge +1Associative MemoryRecurrent Transformers

  3. Fast Weight Attention for Continual Learning

    Aug 27, 2026Yifan Zhang, Steve Ta, Jasper Zhang +8Recurrent Language ModelsMemory-Augmented Language Models