Confidence Self-Calibration for Multi-Label Class-Incremental Learning
This work addresses catastrophic forgetting in multi-label incremental learning, a domain-specific problem for computer vision applications, with incremental improvements over existing methods.
The paper tackles the partial label challenge in Multi-Label Class-Incremental Learning, which causes false-positive errors and catastrophic forgetting, by proposing a Confidence Self-Calibration approach that refines multi-label confidence calibration and achieves new state-of-the-art results on MS-COCO and PASCAL VOC datasets.
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 introduce a class-incremental graph convolutional network that bridges the isolated label spaces by constructing learnable, dynamically extended label relationship graph. Then, for confidence calibration, we present a max-entropy regularization for each multi-label increment, facilitating confidence self-calibration through the penalization of over-confident output distributions. Our approach attains new state-of-the-art results in MLCIL tasks on both MS-COCO and PASCAL VOC datasets, with the calibration of label confidences confirmed through our methodology.