DAM: Dynamic Adapter Merging for Continual Video QA Learning

arXiv:2403.08755v1 Announce Type: cross
Abstract: We present a parameter-efficient method for continual video question-answering (VidQA) learning. Our method, named DAM, uses the proposed Dynamic Adapter Merging to (i) mitigate catastrophic forgetting, (ii) enable efficient adaptation to continually arriving datasets, (iii) handle inputs from unknown datasets during inference, and (iv) enable knowledge sharing across similar dataset domains. Given a set of continually streaming VidQA datasets, we sequentially train dataset-specific adapters for each dataset while freezing the parameters of a large pretrained …

adapter arxiv continual cs.ai cs.cl cs.cv cs.lg dam dynamic merging type video