Confidence Self-Calibration for Multi-Label Class-Incremental Learning

arXiv:2403.12559v1 Announce Type: cross
Abstract: The partial label challenge in Multi-Label Class-Incremental Learning (MLCIL) arises when only the new classes are labeled during training, while past and future labels remain unavailable. This issue leads to a proliferation of false-positive errors due to erroneously high confidence multi-label predictions, exacerbating catastrophic forgetting within the disjoint label space. In this paper, we aim to refine multi-label confidence calibration in MLCIL and propose a Confidence Self-Calibration (CSC) approach. Firstly, for label relationship calibration, we …

abstract arxiv catastrophic forgetting challenge class confidence cs.cv cs.lg errors false false-positive future incremental issue labels leads positive predictions training type