paper-with-me

홈 › Papers

Catastrophic Forgetting is Low-Rank: A Function-Space Theory for Continual Adaptation

2026-06-16 · Ido Nitzan Hidekel, Dan Raviv arxiv

Catastrophic forgetting in continual adaptation is usually studied through parameter drift, replay, or distillation, but these views do not identify which output-space directions are vulnerable. We give a function-space account in the NTK regime: new-task training induces old-task prediction drift through the cross-task kernel, yielding a closed-form predictor for the forgetting vector before any new-task gradient step. In frozen-backbone linear-head PEFT-CL, where the model is linear in the trainable parameters, the predictor is exact up to numerical precision; for nonlinear adapters/full fine-tuning, it is a local NTK approximation. The same expression reveals that forgetting concentrates in a small number of old-task NTK eigenmodes and under frozen linear heads gives a Kronecker scaling rule for the vulnerable rank. These results clarify the relation to prior NTK-overlap theory, explain why parameter-space regularizers can miss output-space interference, and motivate a targeted spectral regularizer.

📄 PDF Abstract BibTeX arXiv:2606.18024

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Subspace Geometry Governs Catastrophic Forgetting in Low-Rank Adaptation

2026-02-10 · Brady Steele arxiv

Low-Rank Adaptation (LoRA) has emerged as a parameter-efficient approach for adapting large pre-trained models, yet its behavior under continual learning remains poorly understood. We present a geometric theory character…

parameter-efficient fine-tuningContinual Learning

Diffusion-based neuromodulation can eliminate catastrophic forgetting in simple neural networks

2017-05-20 · Roby Velez, Jeff Clune

A long-term goal of AI is to produce agents that can learn a diversity of skills throughout their lifetimes and continuously improve those skills via experience. A longstanding obstacle towards that goal is catastrophic …

Diagnostic

On Catastrophic Forgetting in Low-Rank Decomposition-Based Parameter-Efficient Fine-Tuning

2026-03-10 · Muhammad Ahmad, Jingjing Zheng, Yankai Cao arxiv

Parameter-efficient fine-tuning (PEFT) based on low-rank decomposition, such as LoRA, has become a standard for adapting large pretrained models. However, its behavior in sequential learning -- specifically regarding cat…

parameter-efficient fine-tuningContinual Learning

An Empirical Investigation of Catastrophic Forgetting in Gradient-Based Neural Networks

2013-12-21 · Ian J. Goodfellow, Mehdi Mirza, Da Xiao, Aaron Courville 외

Catastrophic forgetting is a problem faced by many machine learning models and algorithms. When trained on one task, then trained on a second task, many machine learning models "forget" how to perform the first task. Thi…

BIG-bench Machine Learning

Complementary Subspace Low-Rank Adaptation of Vision-Language Models for Few-Shot Classification

2025-01-25 · Zhongqi Wang, Jia Dai, Kai Li, Xu Li 외

Vision language model (VLM) has been designed for large scale image-text alignment as a pretrained foundation model. For downstream few shot classification tasks, parameter efficient fine-tuning (PEFT) VLM has gained muc…

Few-Shot Learningparameter-efficient fine-tuning