Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Energy-Structured Low-Rank Adaptation for Continual Learning

About

While orthogonal subspace methods try to mitigate task interference in Continual Learning (CL), they often suffer from energy diffusion across the basis, hindering knowledge compaction and exhausting capacity for future tasks. We observe that output feature drift induced by parameter updates is inherently low-rank, and theoretically prove that preserving parameters along the principal directions of this drift minimizes the output reconstruction error. Motivated by this, we propose \textbf{E}nergy-Concentrated and \textbf{E}nergy-Ordered \textbf{Lo}w-\textbf{R}ank \textbf{A}daptation (E$^2$-LoRA). By explicitly ordering and concentrating knowledge into leading ranks, E$^2$-LoRA frees capacity for subsequent tasks. Furthermore, we design a dynamic rank allocation strategy to balance stability and plasticity by jointly optimizing energy retention and model plasticity. Extensive experiments across multiple benchmarks demonstrate that E$^2$-LoRA achieves state-of-the-art performance.

Longhua Li, Lei Qi, Qi Tian, Xin Geng• 2026

Related benchmarks

TaskDatasetResultRank
Class-incremental learningCIFAR-100--
281
Class-incremental learningImageNet-R
Last Accuracy82.77
147
Class-incremental learningCUB200
Last Accuracy89.77
64
Class-incremental learningCIFAR-100 20 tasks--
58
Class-incremental learningImageNet-R (20 tasks)
Accuracy (20 Tasks)86.37
32
Class-incremental learningCARS 196
Last Accuracy75.82
32
Domain-incremental learningOffice-Home
Last Accuracy88.25
22
Domain-incremental learningDomainNet
Last Step Accuracy69.63
18
Class-incremental learningImageNet-R 50 tasks
Last Accuracy78.58
10
Class-incremental learningCIFAR-100 50 tasks
Accuracy (Last Task)90.7
10
Showing 10 of 10 rows

Other info

Follow for update