Graph Neural Network Based Multi-Instance Learning with Graph Structure Learning

Citations Over TimeTop 25% of 2024 papers

Abstract

Multi-instance learning can make use of the existing weak labeled data to avoid a lot of manual labeling work costs in supervised learning, and has been widely used in text classification, computer-aided diagnosis, and other fields. However, in recent graph neural network based multi-instance learning methods, the quality of the graph is not high, without considering the specific requirements of downstream tasks while constructing the graph. In addition, the network used now is not enough to fully obtain the information of the bags, which limits the performance of the model. In this paper, we propose a graph neural network based multi-instance learning with graph structure learning method, which explicitly introduces the graph structure learning module to improve the quality of the constructed graph, and uses the graph neural network method based on multi-channel and decoupled transformation and propagation to update the node embedding. Experiments on five benchmark datasets demonstrate that our method achieves improvement compared with baselines.

Related Papers

• → Physician-Friendly Machine Learning: A Case Study with Cardiovascular Disease Risk Prediction(2019)71 cited
• → Application of Machine Learning in Animal Disease Analysis and Prediction(2020)26 cited
• → Sentiment Analysis by Using Supervised Machine Learning and Deep Learning Approaches(2020)3 cited
• → Breakdown of Machine Learning Algorithms(2022)1 cited
• → Machine Learning Techniques for the Management of Diseases: A Paper Review(2024)