Rehearsal-free class-incremental learning with LoRA adapters is challenging because low-rank subspaces evolve without geometric control

Rehearsal-free class-incremental learning (CIL) with LoRA adapters remains challenging because the low-rank subspaces updated across tasks evolve without geometric control, causing unstable shared representations and repetitive collapse of task-specific updates into previously occupied directions.

http://arxiv.org/abs/2608.26960v1