Integrating Implicit and Explicit Relational Biases through Graph-Based Multiple Instance Learning: A Case Study in Skin Lesion Diagnosis

2026-08-06

AI Summary

This work addresses the insufficient modeling of structural relationships in skin lesion image classification by proposing a dual-level relational fusion framework. At the implicit level, a convolutional masked autoencoder is employed to learn self-supervised inter-patch relationships, while at the explicit level, a graph attention network leverages multiple graph topologies—such as grid and k-nearest neighbor structures—to facilitate message passing. This approach represents the first effort to jointly exploit implicit self-supervised relations and explicit graph-structured priors, thereby enhancing the model’s structural awareness. The method achieves balanced accuracies of 79.27% and 60.67% on the ISIC-2018 and ISIC-2019 datasets, respectively, significantly outperforming both single-relation modeling strategies and baseline approaches.

Abstract

Relational inductive biases are essential for capturing structural dependencies among data. This study investigates a dual-level relational framework for image classification, bridging the gap between implicit representation learning and explicit structural modelling. We begin by establishing a baseline using an EfficientNetB3 architecture. To move beyond standard convolutional biases, we adopt a patch-based strategy, employing a convolutional masked autoencoder to learn implicit inter-patch relationships through self-supervised reconstruction. We then extend this approach by incorporating explicit relational modelling, organizing the learned embeddings into various graph topologies, including grid-based, random, and k-nearest neighbour structures. Experimental results on the ISIC-2018 and ISIC-2019 skin lesion diagnosis benchmarks show that combining implicit inter-patch modelling with explicit graph-based message passing yields the best performance. On the ISIC-2018 test set, the baseline model achieves a balanced accuracy of 76.17%, which improves to 77.12% with implicit patch-based relational modelling. The fully integrated grid-structured Graph Attention Network further increases performance to 79.27%. Similarly, on ISIC-2019, the implicit approach reaches 59.84% balanced accuracy, while the combination of implicit and explicit modelling yields 60.67%.

Problem

• relational inductive biases
• skin lesion diagnosis
• implicit relational modelling
• explicit graph-based modelling
• image classification

Innovation

• relational inductive bias
• graph-based multiple instance learning
• masked autoencoder
• graph attention network
• patch-based representation