ARMA-C3: A Contrastive ARMA Convolutional Framework for Unsupervised and Semi-supervised Classification
2026-05-25 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors developed ARMA-C3, a new method that helps computers better identify diseases from medical images, even when there are only a few labeled examples. Their approach treats each sample as a point in a network and learns from how these points connect to find useful patterns. They tested it on five different medical datasets, including Alzheimer's and some types of scans, and found it works as well or better than other popular methods, especially when there is little labeled data or imbalanced classes. Their method also works well across different types of medical images.
graph learningcontrastive learningsemi-supervised learningnode classificationmedical imagingAlzheimer's Disease Neuroimaging Initiativeclass imbalancerepresentation learningunsupervised learningcross-modal generalization
Authors
VSS Tejaswi Abburi, Saurabh J. Shigwan, Nitin Kumar
Abstract
In biomedical and neurodegenerative disorders, accurate and early disease identification remains challenging due to the scarcity of labeled data and the complexity of imaging patterns. To address these challenges, we introduce ARMA-C3, a unified unsupervised and semi-supervised graph learning framework for node classification based on contrastive learning and graph-cut regularization to learn structurally meaningful and discriminative representations. By modeling samples or images as graph nodes and exploiting inter-sample relationships, the proposed framework captures subject-level dependencies that conventional machine learning methods typically overlook. We conduct extensive binary classification experiments across five clinically relevant datasets: the Alzheimer's Disease Neuroimaging Initiative (ADNI), the Neuroimaging in Frontotemporal Dementia (NIFD) dataset, and three medical imaging benchmarks (BreastMNIST, PneumoniaMNIST, and a liver ultrasound dataset). Experimental results demonstrate that ARMA-C3 achieves competitive and frequently superior performance compared to classical clustering techniques, state-of-the-art machine learning models, and existing graph-based deep learning approaches across multiple evaluation settings, particularly under limited supervision and severe class imbalance. The proposed framework further demonstrates robust representation learning and strong cross-modal generalization across diverse biomedical imaging modalities.