{"title":"DEEM: A novel approach to semi-supervised and unsupervised image clustering under uncertainty using belief functions and convolutional neural networks","authors":"Loïc Guiziou , Emmanuel Ramasso , Sébastien Thibaud , Sébastien Denneulin","doi":"10.1016/j.ijar.2025.109400","DOIUrl":null,"url":null,"abstract":"<div><div>DEEM (Deep Evidential Encoding of iMages) is a clustering algorithm that combines belief functions with convolutional neural networks in a Siamese-like framework for unsupervised and semi-supervised image clustering. In DEEM, images are mapped to Dempster–Shafer mass functions to quantify uncertainty in cluster membership. Various forms of prior information, including must-link and cannot-link constraints, supervised dissimilarities, and Distance Metric Learning, are incorporated to guide training and improve generalisation. By processing image pairs through shared network weights, DEEM aligns pairwise dissimilarities with the conflict between mass functions, thereby mitigating errors in noisy or incomplete distance matrices. Experiments on MNIST demonstrate that DEEM generalises effectively to unseen data while managing different types of prior knowledge, making it a promising approach for clustering and semi-supervised learning from image data under uncertainty.</div></div>","PeriodicalId":13842,"journal":{"name":"International Journal of Approximate Reasoning","volume":"181 ","pages":"Article 109400"},"PeriodicalIF":3.2000,"publicationDate":"2025-02-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"International Journal of Approximate Reasoning","FirstCategoryId":"94","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0888613X25000416","RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE","Score":null,"Total":0}
引用次数: 0
Abstract
DEEM (Deep Evidential Encoding of iMages) is a clustering algorithm that combines belief functions with convolutional neural networks in a Siamese-like framework for unsupervised and semi-supervised image clustering. In DEEM, images are mapped to Dempster–Shafer mass functions to quantify uncertainty in cluster membership. Various forms of prior information, including must-link and cannot-link constraints, supervised dissimilarities, and Distance Metric Learning, are incorporated to guide training and improve generalisation. By processing image pairs through shared network weights, DEEM aligns pairwise dissimilarities with the conflict between mass functions, thereby mitigating errors in noisy or incomplete distance matrices. Experiments on MNIST demonstrate that DEEM generalises effectively to unseen data while managing different types of prior knowledge, making it a promising approach for clustering and semi-supervised learning from image data under uncertainty.
期刊介绍:
The International Journal of Approximate Reasoning is intended to serve as a forum for the treatment of imprecision and uncertainty in Artificial and Computational Intelligence, covering both the foundations of uncertainty theories, and the design of intelligent systems for scientific and engineering applications. It publishes high-quality research papers describing theoretical developments or innovative applications, as well as review articles on topics of general interest.
Relevant topics include, but are not limited to, probabilistic reasoning and Bayesian networks, imprecise probabilities, random sets, belief functions (Dempster-Shafer theory), possibility theory, fuzzy sets, rough sets, decision theory, non-additive measures and integrals, qualitative reasoning about uncertainty, comparative probability orderings, game-theoretic probability, default reasoning, nonstandard logics, argumentation systems, inconsistency tolerant reasoning, elicitation techniques, philosophical foundations and psychological models of uncertain reasoning.
Domains of application for uncertain reasoning systems include risk analysis and assessment, information retrieval and database design, information fusion, machine learning, data and web mining, computer vision, image and signal processing, intelligent data analysis, statistics, multi-agent systems, etc.