Article
Engineering, Marine
Mei Ling Fam, Dimitrios Konovessis, XuHong He, Lin Seng Ong
Summary: The study proposed using Bayesian Belief Networks (BBN) as part of the risk assessment for decommissioning activities, with initial results showing that database data can be used to structure the BBN and integrating multiple information sources can support risk assessment.
JOURNAL OF OCEAN ENGINEERING AND SCIENCE
(2021)
Article
Engineering, Civil
Roya Meydani, Tommy Giertz, John Leander
Summary: This paper presents a Bayesian decision model for the maintenance planning of water networks, focusing on locating and rehabilitating leakages. The results show that the cost of interventions and probabilities of leakages have a significant influence on the decision-making process.
JOURNAL OF PIPELINE SYSTEMS ENGINEERING AND PRACTICE
(2022)
Article
Radiology, Nuclear Medicine & Medical Imaging
Philip Ciunkiewicz, Michael Roumeliotis, Kailyn Stenhouse, Philip McGeachy, Sarah Quirk, Petra Grendarova, Svetlana Yanushkevich
Summary: The purpose of this analysis is to predict worsening post-treatment normal tissue toxicity in patients undergoing accelerated partial breast irradiation (APBI) therapy and to identify which diagnostic, anatomical, and dosimetric features contribute to these outcomes. A Bayesian network (BN) was constructed using 32 features and outperformed conventional machine learning techniques in predicting tissue toxicity outcomes. The volume of the clinical target volume, gross target volume, and baseline toxicity measurements were found to have the highest feature importance and mutual dependence with normal tissue toxicity.
Article
Automation & Control Systems
Yi Yang, Siyu Huang, Haoran Chen, Meilin Wen, Linhan Gou, Xiao Chen, Liu Wei
Summary: This paper introduces a belief reliability analysis method based on traffic performance margin to assess the reliability of transportation systems. By incorporating uncertainty theory to model epistemic uncertainty, the proposed method considers both stochastic and epistemic uncertainty. It utilizes an uncertain percolation semi Markov model to describe the essential physical characteristics of traffic accidents, and a simulation method to calculate belief reliability based on the traffic failure propagation process.
JOURNAL OF THE FRANKLIN INSTITUTE-ENGINEERING AND APPLIED MATHEMATICS
(2023)
Article
Computer Science, Theory & Methods
Samir Awad, Abdelhamid Malki, Mimoun Malki
Summary: The paper introduces the purpose and method of connecting everyday objects to the Web of Things (WoT), and proposes a probabilistic approach based on Bayesian networks to handle compositions of WoT services with uncertain and correlated data.
Article
Computer Science, Artificial Intelligence
Shaojing Sheng, Xianjie Guo, Kui Yu, Xindong Wu
Summary: This study proposes a novel method for local causal structure learning with missing data, named misLCS. It addresses the issues of low accuracy, low efficiency, and instability in existing algorithms by incorporating iterative data imputation, data subset strategy, and mutual information-based feature selection. Experimental results demonstrate that misLCS outperforms other algorithms in terms of accuracy.
EXPERT SYSTEMS WITH APPLICATIONS
(2024)
Article
Thermodynamics
Puzhe Lan, Dong Han, Xiaoyuan Xu, Zheng Yan, Xijun Ren, Shiwei Xia
Summary: This paper addresses the challenges in state estimation of the coupled electric-gas integrated energy system and proposes a data-driven model using a hybrid deep learning network to improve the accuracy of state estimation.
Article
Multidisciplinary Sciences
Antoine Baker, Indaco Biazzo, Alfredo Braunstein, Giovanni Catania, Luca Dall'Asta, Alessandro Ingrosso, Florent Krzakala, Fabio Mazza, Marc Mezard, Anna Paola Muntoni, Maria Refinetti, Stefano Sarao Mannelli, Lenka Zdeborova
Summary: Research suggests that probabilistic risk estimation can enhance the performance of digital contact tracing, aiding in mitigating the impact of epidemics.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
(2021)
Article
Computer Science, Artificial Intelligence
Jianfeng Qiu, Xiaoqiang Cai, Xingyi Zhang, Fan Cheng, Shenzhi Yuan, Guanglong Fu
Summary: A new evolutionary multi-objective approach (MOEA-PUL) for positive and unlabeled (PU) learning is proposed in this paper, aiming to build a PU classifier without any prior assumption for data distribution and objective functions. The method formulates PU learning as a bi-objective optimization problem using true positive rate (TPR) and a new metric, unlabeled accuracy rate (UAR), as objectives, and demonstrates superiority over existing PU learning algorithms through empirical studies on 12 datasets.
APPLIED SOFT COMPUTING
(2021)
Article
Computer Science, Artificial Intelligence
Tianyu Liu, Jie Lu, Zheng Yan, Guangquan Zhang
Summary: Meta-learning aims to adapt to new tasks quickly and effectively using experience from previous tasks, but the generalization of this adaptation to new tasks remains unclear. The PAC Bayes bound theory provides a theoretical framework for analyzing the generalization performance of meta-learning, with the introduction of new generalization error upper bounds.
Article
Computer Science, Artificial Intelligence
Carlos Ortega Vazquez, Seppe vanden Broucke, Jochen De Weerdt
Summary: Learning from positive and unlabeled data (PU learning) is challenging when there is class imbalance. This paper proposes PU Hellinger Decision Tree (PU-HDT) to directly handle imbalanced PU data sets. Moreover, PU Stratified Hellinger Random Forest (PU-SHRF) is introduced as an ensemble method that outperforms existing PU learning methods for imbalanced data sets in most experimental settings.
Article
Energy & Fuels
Mengda Cao, Tao Zhang, Yajie Liu, Yu Wang, Zhichao Shi
Summary: Simulation tools are essential for stable space mission implementation and satellite operation. They are used to monitor satellite behavior by comparing telemetry values with predicted values. However, retraining the simulation tool based on newly arrived data consumes computing resources and causes time delays. This paper proposes a Bayesian optimization hyperband-optimized incremental learning-based deep belief network (BOHB-ILDBN) to accurately and quickly predict satellite behavior. The model is tested and verified using telemetry data from the CBERS-4A satellite.
JOURNAL OF ENERGY STORAGE
(2023)
Article
Computer Science, Artificial Intelligence
Lianmeng Jiao, Haoyu Yang, Feng Wang, Zhun-ga Liu, Quan Pan
Summary: A decision tree-based evidential clustering algorithm is proposed to improve the interpretability of the resulting cluster assignments. The algorithm uses the paths from the root node to leaf nodes to achieve the interpretability of each cluster. Experimental results show that the proposed algorithm performs well compared to representative fuzzy, evidential, or decision tree-based clustering algorithms.
PATTERN RECOGNITION
(2023)
Article
Chemistry, Multidisciplinary
Lea Vojkovic, Ana Kuzmanic Skelin, Djani Mohovic, Damir Zec
Summary: An integrative approach using Bayesian networks is proposed for maritime accident risk factor assessment, with a novel quantitative assessor introduced. The method is showcased in the context of small passenger ship grounding cases, demonstrating its capability for causal factor identification.
APPLIED SCIENCES-BASEL
(2021)
Article
Engineering, Civil
Mahmood Ahmad, Xiao-Wei Tang, Jiang-Nan Qiu, Feezan Ahmad, Wen-Jing Gu
Summary: This study evaluated the seismic soil liquefaction potential using four machine learning algorithms, showing that the K2 and TAN Bayes models outperformed Tabu search and HC models. Sensitivity analysis indicated that cone tip resistance and vertical effective stress are the most sensitive factors, while mean grain size is the least sensitive.
FRONTIERS OF STRUCTURAL AND CIVIL ENGINEERING
(2021)