---
title: "Towards Robust Incremental Learning Under Ambiguous Supervision"
authors:
  - "Rui Wang"
  - "Mingxuan Xia"
  - "Haobo Wang"
  - "Lei Feng"
  - "Junbo Zhao"
  - "Gang Chen"
  - "Chang Yao"
doi: "10.24963/ijcai.2025/712"
url: "https://ijcai.org/proceedings/2025/712"
pdf: "https://ijcai.org/proceedings/2025/0712.pdf"
venue: "Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence"
year: 2025
pages: "6397-6405"
track: "Main Track"
keywords:
  - "Machine Learning: ML: Weakly supervised learning"
  - "Machine Learning: ML: Semi-supervised learning"
---

# Towards Robust Incremental Learning Under Ambiguous Supervision

Rui Wang, Mingxuan Xia, Haobo Wang, Lei Feng, Junbo Zhao, Gang Chen, Chang Yao

*Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence*, pages 6397-6405. DOI: [10.24963/ijcai.2025/712](https://doi.org/10.24963/ijcai.2025/712)

## Abstract

Traditional Incremental Learning (IL) targets to handle sequential fully-supervised learning problems where novel classes emerge from time to time. However, due to inherent annotation uncertainty and ambiguity, collecting high-quality annotated data in a dynamic learning system can be extremely expensive. To mitigate this problem, we propose a novel weakly-supervised learning paradigm called Incremental Partial Label Learning (IPLL), where the sequentially arrived data relate to a set of candidate labels rather than the ground truth. Technically, we develop the Prototype-Guided Disambiguation and Replay Algorithm (PGDR) which leverages the class prototypes as a proxy to mitigate two intertwined challenges in IPLL, i.e., label ambiguity and catastrophic forgetting. To handle the former, PGDR encapsulates a momentum-based pseudo-labeling algorithm along with prototype-guided initialization, resulting in a balanced perception of classes. To alleviate forgetting, we develop a memory replay technique that collects well-disambiguated samples while maintaining representativeness and diversity. By jointly distilling knowledge from curated memory data, our framework exhibits a great disambiguation ability for samples of new tasks and achieves less forgetting of knowledge. Extensive experiments demonstrate that PGDR achieves superior performance over the baselines in the IPLL task.