# Rebalancing Multi-Label Class-Incremental Learning

arXiv: https://arxiv.org/abs/2408.12161 · HTML: https://mlsanity.com/paper/2408.12161/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 19 of its 32 references that are in the map.
- First posted on arXiv 22 Aug 2024.

## Builds on (most cited first, 19 of 19)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [Overcoming catastrophic forgetting in neural networks](https://mlsanity.com/paper/1612.00796/) (2016)
- [Learning without Forgetting](https://mlsanity.com/paper/1606.09282/) (2016)
- [Experience Replay for Continual Learning](https://mlsanity.com/paper/1811.11682/) (2018)
- [Learning to Prompt for Continual Learning](https://mlsanity.com/paper/2112.08654/) (2021)
- [DyTox: Transformers for Continual Learning with DYnamic TOken eXpansion](https://mlsanity.com/paper/2111.11326/) (2021)
- [Continual Prototype Evolution: Learning Online from Non-Stationary Data Streams](https://mlsanity.com/paper/2009.00919/) (2020)
- [TResNet: High Performance GPU-Dedicated Architecture](https://mlsanity.com/paper/2003.13630/) (2020)
- [Imbalanced Continual Learning with Partitioning Reservoir Sampling](https://mlsanity.com/paper/2009.03632/) (2020)
- [SSUL: Semantic Segmentation with Unknown Label for Exemplar-based Class-Incremental Learning](https://mlsanity.com/paper/2106.11562/) (2021)
- [Exploring Example Influence in Continual Learning](https://mlsanity.com/paper/2209.12241/) (2022)
- [Incremental Learning via Rate Reduction](https://mlsanity.com/paper/2011.14593/) (2020)
- [Multi-Domain Multi-Task Rehearsal for Lifelong Learning](https://mlsanity.com/paper/2012.07236/) (2020)
- [Acknowledging the Unknown for Multi-label Learning with Single Positive Labels](https://mlsanity.com/paper/2203.16219/) (2022)
- [AGCN: Augmented Graph Convolutional Network for Lifelong Multi-label Image Recognition](https://mlsanity.com/paper/2203.05534/) (2022)
- [Optimizing Class Distribution in Memory for Multi-Label Online Continual Learning](https://mlsanity.com/paper/2209.11469/) (2022)
- [D3Former: Debiased Dual Distilled Transformer for Incremental Learning](https://mlsanity.com/paper/2208.00777/) (2022)

## Built on by (most cited first, 2 of 2)

- [Class-Independent Increment: An Efficient Approach for Multi-label Class-Incremental Learning](https://mlsanity.com/paper/2503.00515/) (2025)
- [P2L-CA: An Effective Parameter Tuning Framework for Rehearsal-Free Multi-Label Class-Incremental Learning](https://mlsanity.com/paper/2601.12714/) (2026)

## Abstract (arXiv)

Multi-label class-incremental learning (MLCIL) is essential for real-world multi-label applications, allowing models to learn new labels while retaining previously learned knowledge continuously. However, recent MLCIL approaches can only achieve suboptimal performance due to the oversight of the positive-negative imbalance problem, which manifests at both the label and loss levels because of the task-level partial label issue. The imbalance at the label level arises from the substantial absence of negative labels, while the imbalance at the loss level stems from the asymmetric contributions of the positive and negative loss parts to the optimization. To address the issue above, we propose a Rebalance framework for both the Loss and Label levels (RebLL), which integrates two key modules: asymmetric knowledge distillation (AKD) and online relabeling (OR). AKD is proposed to rebalance at the loss level by emphasizing the negative label learning in classification loss and down-weighting the contribution of overconfident predictions in distillation loss. OR is designed for label rebalance, which restores the original class distribution in memory by online relabeling the missing classes. Our comprehensive experiments on the PASCAL VOC and MS-COCO datasets demonstrate that this rebalancing strategy significantly improves performance, achieving new state-of-the-art results even with a vanilla CNN backbone.