Reinforced GNNs for Multiple Instance Learning

Xusheng Zhao

doi:10.1109/TNNLS.2024.3392575

Reinforced GNNs for Multiple Instance Learning

2024-01-01

0

PRE

AI

DOI：10.1109/TNNLS.2024.3392575

Original

Original request for help

Save

Abstract

En 中文

Multiple instance learning (MIL) trains models from bags of instances, where each bag contains multiple instances, and only bag-level labels are available for supervision. The application of graph neural networks (GNNs) in capturing intrabag topology effectively improves MIL. Existing GNNs usually require filtering low-confidence edges among instances and adapting graph neural architectures to new bag structures. However, such asynchronous adjustments to structure and architecture are tedious and ignore their correlations. To tackle these issues, we propose a reinforced GNN framework for MIL (RGMIL), pioneering the exploitation of multiagent deep reinforcement learning (MADRL) in MIL tasks. MADRL enables the flexible definition or extension of factors that influence bag graphs or GNNs and provides synchronous control over them. Moreover, MADRL explores structure-to-architecture correlations while automating adjustments. Experimental results on multiple MIL datasets demonstrate that RGMIL achieves the best performance with excellent explainability.

Keywords:

Graph neural networks

Filtering

Correlation

Training

Supervised learning

Indexes

Electronic mail

Deep reinforcement learning (RL)

graph neural network (GNN)

multiple instance learning (MIL)

neural architecture search

Reinforced GNNs for Multiple Instance Learning

Abstract

Journal

Organization