Random Amalgamation of Adapters for Flatter Loss Landscapes: Towards Class-Incremental Learning with Better Stability

Yao Deng; Xiang Xiang; Jiaqi Gui

AAAI 2026technical0 citations

Random Amalgamation of Adapters for Flatter Loss Landscapes: Towards Class-Incremental Learning with Better Stability

Abstract

Class-incremental learning (CIL) enables models to continuously learn from streaming data while mitigating catastrophic forgetting of prior knowledge. Our research reveals that the CIL performance of pre-trained models (PTMs) varies significantly across different datasets, a phenomenon underexplored in existing studies. Through visualization, we observe that flatter loss landscapes correlate with superior CIL performance. This insight motivates us to enhance PTMs

BibTeX

@inproceedings{aaai2026_randomamalgamati,
  title = {Random Amalgamation of Adapters for Flatter Loss Landscapes: Towards Class-Incremental Learning with Better Stability},
  author = {Yao Deng and Xiang Xiang and Jiaqi Gui},
  booktitle = {AAAI 2026},
  year = {2026}
}