Semantically-Shifted Incremental Adapter-Tuning is A Continual ViTransformer

arXiv:2403.19979v1 Announce Type: cross
Abstract: Class-incremental learning (CIL) aims to enable models to continuously learn new classes while overcoming catastrophic forgetting. The introduction of pre-trained models has brought new tuning paradigms to CIL. In this paper, we revisit different parameter-efficient tuning (PET) methods within the context of continual learning. We observe that adapter tuning demonstrates superiority over prompt-based methods, even without parameter expansion in each learning session. Motivated by this, we propose incrementally tuning the shared adapter without imposing parameter …

abstract adapter arxiv catastrophic forgetting class context continual cs.ai cs.cv cs.lg incremental introduction learn observe paper pet pre-trained models type