The theory of belief functions, also referred to as evidence theory or Dempster-Shafer theory, was first introduced by Arthur P. Dempster in the context of statistical inference, and was later developed by Glenn Shafer as a general framework for modelling epistemic uncertainty. These early contributions have been the starting points of many important developments, including the transferable belief model and the theory of hints. The theory of belief functions is now well established as a general framework for reasoning with uncertainty, and has well understood connections to other frameworks such as probability, possibility and imprecise probability theories. It has been applied in diverse areas such as machine learning, information fusion and risk analysis.
The BELIEF conferences, sponsored by the Belief Functions and Applications Society, are dedicated to the confrontation of ideas, the reporting of recent achievements and the presentation of the wide range of applications of this theory. The first edition of this conference series was held in Brest, France, in 2010. Later editions were held in Compiègne, France in 2012, Oxford, UK in 2014, Prague, Czech Republic in 2016, and again in Compiègne, France in 2018. The Sixth International Conference on Belief Functions (BELIEF 2021) will be located in Shanghai, China, on October 15-17, 2021, together with the 2021 International Conference on Cognitive analytics, Granular computing, and Three-way decisions (CCGT). It will be held both onsite and online due to the COVID-19 situation (see Venue and Registration below for details).
The expected length of papers is no longer than 10 pages, references included, that should present original contributions with significant results. Springer encourages authors to include their ORCIDs in their papers. In addition, the corresponding author of each accepted paper, acting on behalf of all of the authors of that paper, will have to complete and sign a Consent-to-Publish form. The corresponding author signing the copyright form should match the corresponding author marked on the paper. Once the files have been sent to Springer, changes relating to the authorship of the papers cannot be made.
Original contributions are solicited on theoretical aspects including, but not limited to
as well as on applications to various areas including, but not limited to
Authors of selected papers from the BELIEF 2021 conference will be invited to submit extended versions of their papers for possible inclusion in a special issue of the International Journal of Approximate Reasoning.
The program of this edition of the BELIEF conference will include tutorials from experts on belief functions and their applications.
To accommodate for the uncertainties surrounding travel possibilities due to the COVID-19 pandemic, participants have two options to attend the conference: either online or onsite.
The onsite event will take place at the Gu Cun Park Hotel (No.4788 Hu Tai Road, Baoshan District, Shanghai). The on-site participants can arrive to hotel by any ways (such as taxi, bus or Line 7 subway).
To facilitate offline communication, a Wechat group for on-site participants have also been created.
Online participants will be able to join and participate live to the onsite event via this link (R1 platform). We expect most keynote talks to be given onsite (and tutorials to be given online). The conference will also be recorded and the recording made available on the R1 platform of the conference: https://researchers.one/conferences/belief2021.
The registration fee for the onsite participants includes the following items:
Online participants benefit from a discounted registration fee, which includes the following items:
|Additional ticket for gala dinner
|Additional ticket for welcome reception
On-site and online participants have to register using link (R1 platform) and have to pay the registration fees described on this page (100€ for a non-student participant, 50€ for a student participant). Each on-site participant will have to pay the registration fee 130€ for all the on-site conference services.
(Shanghai time UTC+8)
BELIEF proceedings can be found here SpringerLink - Free access granted until end of November 2021.
|Friday October 15
|Session 1 (Classification)
|Session 2 (Information Fusion)
|Session 3 (Statistical Inference and Learning)
|Saturday October 16
| Keynote by Van Nam Huynh
Machine Learning coupled with Evidential Reasoning for User Preference
Chair: Xiaodong Yue
| Keynote by Chunlai Zhou
Basic Utility Theory for Belief Functions
Chair: Zhunga Liu
|Session 4 (Elicitation)
|Session 5 (Deep Learning)
|Session 6 (Conflict, inconsistency and specificity)
|BFAS General Assembly
|Sunday October 17
| Keynote by Deqiang Han
Learning-based Modelized Methods for Evidence Combination
Chair: Zhunga Liu
| Keynote by Zengjing Chen
A Central Limit Theorem for Sets of Probability Measures
Chair: Xiaodong Yue
|Session 7 (Clustering)
|Session 8 (Transfer Learning)
|Session 9 (Algorithms and Computation)
Instruction for oral presentation
Each presentation is scheduled to be 15 minutes long, including questions. Instruction to chairman is to leave 10 minutes for the presentation itself and 5 minutes for questions/discussions
|Session 1 (Classification) (chair: Liyao Ma)
| Improving Micro-Extended Belief Rule-Based System using Activation Factor for Classification Problems
Long-Hao Yang, Jun Liu, Ying-Ming Wang, Hui Wang and Luis Martínez
| Orbit Classification for Prediction Based on Evidential Reasoning and Belief Rule Base
Chao Sun, Xiaoxia Han, Wei He and Hailong Zhu
| Imbalance Data Classification Based on Belief Function Theory
Jiawei Niu and Zhunga Liu
| A Classification Tree Method Based on Belief Entropy for Evidential Data
Kangkai Gao, Liyao Ma and Yong Wang
|Session 2 (Information Fusion) (chair: Frédéric Pichon)
| A New Multi-Source Information Fusion Method Based on Belief Divergence Measure and the Negation of Basic Probability Assignment
Hongfei Wang, Wen Jiang, Xinyang Deng and Jie Geng
| Improving an Evidential Source of Information Using Contextual
Corrections Depending on Partial Decisions
Siti Mutmainah, Samir Hachour, Frédéric Pichon and David Mercier
|Session 3 (Statistical Inference and Learning) (chair: Ryan Martin)
| Entropy-based Learning of Compositional Models from Data
Radim Jiroušek, Václav Kratochvíl and Prakash P. Shenoy
| Approximately Valid and Model-Free Possibilistic Inference
Leonardo Cella and Ryan Martin
| Towards a Theory of Valid Inferential Models with Partial Prior
| Ensemble Learning Based on Evidential Reasoning Rule with a New Weight Calculation Method
Cong Xu, Zhi-Jie Zhou, Wei He, Hailong Zhu and Yan-Zi Gao
|Session 4 (Elicitation) (chair: Arnaud Martin)
| Validation of Smets’ Hypothesis in the Crowdsourcing Environment
Constance Thierry, Arnaud Martin, Jean-Christophe Dubois and Yolande Le Gall
| Quantifying Confidence of Safety Cases with Belief Functions
Yassir Idmessaoud, Didier Dubois and Jérémie Guiochet
|Session 5 (Deep Learning) (chair: Thierry Denœux)
| Evidential Segmentation of 3D PET/CT Images
Ling Huang, Su Ruan, Pierre Decazes and Thierry Denœux
| Fusion of Evidential CNN Classifiers for Image Classification
Zheng Tong, Philippe Xu and Thierry Denœux
| Multi-branch Recurrent Attention Convolutional Neural Network with Evidence Theory for Fine-grained Image Classification
Zhikang Xu, Bofeng Zhang, Haijie Fu, Xiaodong Yue and Ying Lv
| Deep Evidential Fusion Network for Image Classification
Shaoxun Xu, Yufei Chen, Chao Ma and Xiaodong Yue
|Session 6 (Conflict, inconsistency and specificity) (chair: Anne-Laure Jousselme)
| Conflict Measure of Belief Functions with Blurred Focal Elements on the Real Line
| Logical and Evidential Inconsistencies Meet: First Steps
Nadia Ben Abdallah, Sébastien Destercke, Anne-Laure Jousselme and Frédéric Pichon
| A Note About Entropy and Inconsistency in Evidence Theory
Anne-Laure Jousselme, Frédéric Pichon, Nadia Ben Abdallah and Sébastien Destercke
| An Extension of Specificity-Based Approximations to Other Belief Function Relations
Tekwa Tedjini, Sohaib Afifi, Frédéric Pichon and Éric Lefèvre
|Session 7 (Clustering) (chair: Kuang Zhou)
| Fast Unfolding of Credal Partitions in Evidential Clustering
Zuowei Zhang, Arnaud Martin, Zhunga Liu, Kuang Zhou and Yiru Zhang
| Credal Clustering for Imbalanced Data
Zuowei Zhang, Zhunga Liu, Kuang Zhou, Arnaud Martin and Yiru Zhang
| Evidential Weighted Multi-View Clustering
Kuang Zhou, Mei Guo and Ming Jiang
| Unequal Singleton Pair Distance for Evidential Preference Clustering
Yiru Zhang and Arnaud Martin
|Session 8 (Transfer Learning) (chair: Lianmeng Jiao)
| Transfer Evidential C-means Clustering
Lianmeng Jiao, Feng Wang and Quan Pan
| Evidential Clustering Based on Transfer Learning
Kuang Zhou, Mei Guo and Arnaud Martin
| Ensemble of Adapters for Transfer Learning Based on Evidence Theory
Ying Lv, Bofeng Zhang, Xiaodong Yue, Zhikang Xu and Wei Liu
|Session 9 (Algorithms and Computation) (chair: Juan Jesús Salamanca)
| Discussions on the Connectedness of a Random Closed Set
Juan Jesús Salamanca
| An Efficient Computation of Dempster-Shafer Theory of Evidence Based on Native GPU Implementation
Noelia Rico, Luigi Troiano and Irene Díaz
| QLEN: Quantum-Like Evidential Networks for Predicting the Decision in Prisoner’s Dilemma
Jixiang Deng and Yong Deng
Professor Van Nam Huynh, Japan Advanced Institute of Science and Technology, Japan.
Title: Machine Learning coupled with Evidential Reasoning for User Preference
Abstract: Inferring user preferences from short texts generated by users on social platforms has a variety of applications in web-based decision support systems such as recommender systems and personalized marketing systems. Developing an efficient solution to this problem is still challenging due to difficulty in handling short texts and dynamic change of user preferences over time. In this talk, we will present a novel framework that tackles these challenges by combining advanced Machine Learning techniques for concept learning and Dempster-Shafer theory (DST) for reasoning and fusion to effectively infer user preferences. Two instances of the proposed framework will be demonstrated with experimental results and analysis that show the effectiveness and practicality of the developed methods.
Ass. Professor Chunlai Zhou, Renmin University, China.
Title: Basic Utility Theory for Belief Functions
Abstract: I will talk about a basic utility theory for belief functions which is common ground for different decision theories in Dempster-Shafer theory where the completeness requirement is dropped. The resulting preference relation is represented by subjective expectation of sets of utilities whose ordering is based on an ordering of outcome sets derived from a logical decision theory for complete ignorance. Moreover, we explore the preference aggregation problem within the utility theory and generalize some results by Harsanyi and Mongin to the setting of belief functions.
Professor Deqiang Han, Xi'an Jiaotong University, China.
Title: Learning-based Modelized Methods for Evidence Combination
Abstract: Evidence combination is typical uncertainty reasoning or information fusion in the theory of belief functions, which combines bodies of evidence stemming from different information sources. In traditional applications of evidence combination (e.g., pattern classification), given a sample, the basic belief assignments (BBAs) of different information sources are generated first, and then they are combined by a rule, e.g., Dempster's rule. We propose a modelized method for evidence combination. By just inputting the sample into the learned model of combination, a “combined” BBA is obtained. That is, it does not need to generate multiple BBAs for each sample for the combination. In our proposed modelized combination, one can generate different combination models with different combination rules. Experimental results and related analyses validate the related rationality and efficiency.
Professor Zengjing Chen, Shandong University, China.
Title: A Central Limit Theorem for Sets of Probability Measures
Abstract: We prove a central limit theorem for a sequence of random variables whose means are ambiguous and vary in an unstructured way. Their joint distribution is described by a set of measures. The limit is (not the normal distribution and is) defined by a backward stochastic differential equation that can be interpreted as modeling an ambiguous continuous-time random walk.