---
title: "GroupMIL: Semantic Group Based Multiple Instance Learning for Whole Slide Image Analysing"
authors:
  - "Zhao Yao"
  - "Zhenmi Xie"
  - "Mengxin Tian"
  - "Guoqing Wu"
  - "Yaonan Wang"
  - "Min Liu"
doi: "10.24963/ijcai.2026/778"
url: "https://ijcai.org/proceedings/2026/778"
pdf: "https://ijcai.org/proceedings/2026/0778.pdf"
venue: "Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence"
year: 2026
pages: "6993-7000"
track: "AI and Health"
keywords:
  - "Medical diagnosis: Medical diagnosis"
  - "Health data mining: Health data mining"
  - "Medical imaging: Medical imaging"
---

# GroupMIL: Semantic Group Based Multiple Instance Learning for Whole Slide Image Analysing

Zhao Yao, Zhenmi Xie, Mengxin Tian, Guoqing Wu, Yaonan Wang, Min Liu

*Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence*, pages 6993-7000. DOI: [10.24963/ijcai.2026/778](https://doi.org/10.24963/ijcai.2026/778)

## Abstract

Whole Slide Image (WSI) analysis faces challenges due to gigapixel resolutions and slide-level weak supervision. Multiple Instance Learning (MIL) serves as a pivotal method for this task. However, existing MIL frameworks often fail to exploit the inherent redundancy of tissue patterns or the semantic coherence among similar patches within a WSI. We propose GroupMIL, a novel framework that introduces a differentiable grouping mechanism into the MIL framework. This approach enables the automatic emergence of semantic segments using only slide-level labels. We specifically introduce a multi-stage grouping block and a hierarchical aggregator, which progressively fuse features within and across groups to construct a robust slide-level representation. Extensive experiments on multiple public datasets across cancer subtyping and survival prediction tasks demonstrate that GroupMIL consistently surpasses state-of-the-art performance.