Class-Incremental Learning (CIL) faces challenges with imbalanced data distribution, leading to skewed gradient updates and catastrophic forgetting. The proposed method reweights gradients to balance optimization and mitigate forgetting. Distribution-aware knowledge distillation loss aligns output logits with lost training data distribution. Experimental results show consistent improvements across various datasets and evaluation protocols.
In eine andere Sprache
aus dem Quellinhalt
arxiv.org
Wichtige Erkenntnisse aus
by Jiangpeng He... um arxiv.org 02-29-2024
https://arxiv.org/pdf/2402.18528.pdfTiefere Fragen