Article
Physics, Multidisciplinary
Shouta Sugahara, Maomi Ueno
Summary: Previous research has shown that the classification accuracies of Bayesian networks obtained by maximizing the conditional log likelihood were higher than those obtained by maximizing the marginal likelihood. However, in cases with small sample sizes and a class variable with multiple parents, the accuracies of exact learning with ML were significantly lower. Introducing an exact learning augmented naive Bayes classifier improved the situation and guaranteed similar class posterior estimation as exact learning Bayesian networks.
Article
Physics, Nuclear
Yifan Liu, Chen Su, Jian Liu, Pawel Danielewicz, Chang Xu, Zhongzhou Ren
Summary: The iNBP classifier is proposed to study nuclear masses by refining results of sophisticated nuclear models. It treats the prediction as a classification problem, with impressive improvements on global descriptions and robust extrapolating capabilities. The method can be applied to predict nuclear masses in unknown regions.
Article
Chemistry, Multidisciplinary
Guiliang Ou, Yulin He, Philippe Fournier-Viger, Joshua Zhexue Huang
Summary: This paper proposes an improved approach for constructing NBC called MAF-NBC, which overcomes the limitations of the NBC through a mixed-attribute fusion mechanism and an improved autoencoder neural network. Experimental results demonstrate that MAF-NBC outperforms eight state-of-the-art Bayesian algorithms in classification performance.
APPLIED SCIENCES-BASEL
(2022)
Article
Computer Science, Artificial Intelligence
Tzu-Tsung Wong, Hsing-Chen Tsai
Summary: This study proposes methods to find noninformative generalized Dirichlet priors for multinomial naive Bayesian classifier to greatly improve its performance on high-dimensional imbalanced data.
The methods are tested on seven high-dimensional imbalanced datasets, demonstrating that the multinomial naive Bayesian classifier with generalized Dirichlet priors can significantly outperform not only the original classifier, but also random forest and Ripper algorithm.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Computer Science, Artificial Intelligence
Hongjia Ren, Qiulin Guo
Summary: This paper introduces a Tree Augmented Naive Bayes classifier (TAN) widely used in machine learning and data mining. To enhance the flexibility and classification performance of TAN, a Flexible Tree Augmented Naive Bayes classifier (FTAN) is proposed, which measures attribute dependencies using mutual information contribution rate and filters out weak dependencies through a flexible threshold adjustment. Experimental results on UCI datasets demonstrate the considerable advantages of FTAN in terms of 0-1 loss and class probability root mean square error. The application of FTAN in predicting the favorable distribution area for remaining oil and gas resources in the Junggar Basin shows its effectiveness and superiority, providing a decision-making basis for optimizing drilling strategies and exploration targets.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Engineering, Geological
Laura Cataldi, Lara Tiberi, Giovanni Costa
Summary: The study investigates the relationship between seismic intensity and ground motion parameters, proposing a new approach for handling intensity using Gaussian Naive Bayes classifiers. By expanding and resampling the existing database, new regression relations and probability distributions are estimated. Results show that GNB models outperform GMICEs in terms of performance on unseen data and classification scores.
BULLETIN OF EARTHQUAKE ENGINEERING
(2021)
Article
Construction & Building Technology
Peipei Wang, Kun Wang, Yunhan Huang, Jianguo Zhu, Peter Fenn, Yi Zhang
Summary: The aim of this paper is to understand the formation mechanism of payment issues faced by contractors in China's construction industry and establish a predictive model. The critical factors were identified and categorized through literature review and a Delphi survey. A focus group was conducted to understand the roles played by each category of factors. The associations among the factors were deduced through logical deduction and a Bayesian belief network parameter learning was used to quantify the model structure.
JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT
(2023)
Article
Ecology
N. S. Abinaya, D. Susan, Rakesh S. Kumar
Summary: A fish classification technique using deep learning networks and naive Bayesian fusion is proposed in this work to address challenges in aquaculture industries. By adjusting orientation, multi-stage optimization, and training deep learning networks, high accuracy classification results are achieved.
ECOLOGICAL INFORMATICS
(2021)
Article
Agriculture, Multidisciplinary
Chengquan Zhou, Guijun Yang, Dong Liang, Jun Hu, Hao Yang, Jibo Yue, Ruirui Yan, Liang Han, Linsheng Huang, Lijun Xu
Summary: A machine learning method based on multiple features was proposed to accurately identify and evaluate black point disease in wheat kernels through segmentation, feature extraction, and classification model. The experimental results demonstrated that this method showed high accuracy and effectiveness in identifying and evaluating the incidence of black point disease.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2021)
Article
Environmental Sciences
Hyeok Kim, Dong-Hyeok Park, Jae-Hyun Ahn, Tae-Woong Kim
Summary: In recent years, the frequency and intensity of drought in South Korea have increased due to climate change, leading to increasing socio-economic damage. The concentration of precipitation in the summer and water shortage caused by population growth and urbanization contribute to the severity of droughts. Comprehensive assessment of multiple factors is essential for planning effective drought mitigation strategies.
Article
Biochemistry & Molecular Biology
Jie Gong, Chong Shen, Meng Xiao, Huifang Zhang, Fei Zhao, Jiangzhong Zhang, Di Xiao
Summary: By utilizing MALDI-TOF MS technology and a modified naive Bayesian classifier, this study successfully identified C. krusei and C. auris in mixed samples of fungal coinfections with high sensitivity and specificity. It provides a potential method for identifying intrinsically resistant fungal species and highlights the application of MALDI-TOF MS for analyzing coinfections of different species.
Article
Health Care Sciences & Services
Kimiya Gohari, Anoshirvan Kazemnejad, Marjan Mohammadi, Farzad Eskandari, Samaneh Saberi, Maryam Esmaieli, Ali Sheidaei
Summary: This study proposes a novel approach to enhancing the Naive Bayes classifier by introducing a latent variable as the parent of the attributes, resulting in the NB-BLCA model. By incorporating the latent variable, the model aims to capture complex dependencies among the attributes and improve the overall performance of the classifier.
BMC MEDICAL RESEARCH METHODOLOGY
(2023)
Article
Engineering, Multidisciplinary
Huimin Wang, Zhaojun Steven Li
Summary: A tree augmented naive Bayesian (TAN) classifier is adopted for transient stability assessment of power systems, with the AdaBoost algorithm used for performance improvement. Eight attributes reflecting system stability are selected for the classifier, and a class-attribute interdependence maximization (CAIM) algorithm is used for attribute discretization. Evaluation indicators including Kappa, AUC, F-1 score, and average evaluation indicator show significant improvement in classification performance.
PROCEEDINGS OF THE INSTITUTION OF MECHANICAL ENGINEERS PART O-JOURNAL OF RISK AND RELIABILITY
(2022)
Article
Psychiatry
Haewon Byeon
Summary: This study analyzed epidemiological survey data representing South Korean female older adults living alone to understand the degree of their depressive disorders and factors affecting these disorders. Machine learning techniques were used to identify the main risk factors, and a nomogram was developed to help primary physicians identify high-risk groups for depressive disorders. The study found that stress perception, subjective health, fatty acid intake, sitting time, and sleep hours were major variables related to depressive disorders in female older adults living alone.
FRONTIERS IN PSYCHIATRY
(2022)
Article
Optics
Zhou Mei, Chang Jianhua, Chen Sicheng, Meng Yuanyuan, Dai Tengfei
Summary: The study proposed an aerosol classification model based on aerosol optical inversion data using a naive Bayesian classifier. Experimental results showed that the model achieved high consistency with traditional classification algorithms, providing ground data support for aerosol inversion by remote sensing equipment.
ACTA OPTICA SINICA
(2022)
Article
Environmental Sciences
Chung-Chi Chen, Chun-Yi Lu, Sen Jan, Chih-hao Hsieh, Chih-Ching Chung
Summary: The intensity of coastal uplift along the subtropical region of eastern Taiwan is positively correlated to the flow volume transport of the Kuroshio current. The uplifted nutrients enhance the growth of phytoplankton and may support more energy transfer to higher trophic levels in this oligotrophic ecosystem.
FRONTIERS IN MARINE SCIENCE
(2022)
Article
Environmental Sciences
Yin-Zheng Lai, Chih-Wei Tu, Chih-hao Hsieh, Chia-Ying Ko
Summary: Environmental and climatic changes are expected to redistribute species, altering the strengths of species interaction networks. One way to infer species interaction networks is by analyzing their geographical overlaps, which provides indices of species interdependence. Integrating MSR and MSS further allows us to assess community coexistence stability and structure, with a stronger negative relationship between MSR and MSS within a community suggesting a more stable community.
FRONTIERS IN MARINE SCIENCE
(2022)
Article
Oceanography
Ariana Chih-Hsien Liu, Feng-Hsun Chang, Jinny Wu Yang, Hiroaki Saito, Yu Umezawa, Chung-Chi Chen, Sen Jan, Chih-hao Hsieh
Summary: This study investigated the diversity and taxonomic composition of free-living bacterioplankton along the Kuroshio. The results showed that temperature, phosphate concentration, and chlorophyll-a concentration were important factors affecting the diversity and composition of bacterioplankton.
DEEP-SEA RESEARCH PART I-OCEANOGRAPHIC RESEARCH PAPERS
(2022)
Article
Microbiology
Wan-Hsuan Cheng, Chih-hao Hsieh, Chun-Wei Chang, Fuh-Kwo Shiah, Takeshi Miki
Summary: This study quantitatively examines the functional redundancy of ecosystem functions in a freshwater system. The findings suggest that taxon-based functional specificity is a better predictor of functional redundancy than substrate-based functional specificity. These results provide a framework for predicting the consequences of biodiversity loss on ecosystem functioning.
FEMS MICROBIOLOGY ECOLOGY
(2022)
Correction
Multidisciplinary Sciences
Masayuki Ushio, Chih-hao Hsieh, Reiji Masuda, Ethan R. Deyle, Hao Ye, Chun-Wei Chang, George Sugihara, Michio Kondoh
Letter
Multidisciplinary Sciences
Chun-Wei Chang, Stephan B. Munch, Chih-hao Hsieh
NATURE COMMUNICATIONS
(2022)
Article
Multidisciplinary Sciences
Chun-Wei Chang, Takeshi Miki, Hao Ye, Sami Souissi, Rita Adrian, Orlane Anneville, Helen Agasild, Syuhei Ban, Yaron Be'eri-Shlevin, Yin-Ru Chiang, Heidrun Feuchtmayr, Gideon Gal, Satoshi Ichise, Maiko Kagami, Michio Kumagai, Xin Liu, Shin-Ichiro S. Matsuzaki, Marina M. Manca, Peeter Noges, Roberta Piscia, Michela Rogora, Fuh-Kwo Shiah, Stephen J. Thackeray, Claire E. Widdicombe, Jiunn-Tzong Wu, Tamar Zohary, Chih-hao Hsieh
Summary: Understanding the causal links and feedbacks among biodiversity, ecosystem functioning, and environmental factors is challenging due to their complex interactions. Through empirical dynamic modeling, this study reveals macroecological patterns and varying feedback strengths in aquatic ecosystems. The findings highlight the importance of considering networks in future ecosystem management.
NATURE COMMUNICATIONS
(2022)
Article
Environmental Sciences
Gael Dur, Xin Liu, Yoichiro Sakai, Chih-hao Hsieh, Syuhei Ban, Sami Souissi
Summary: The seasonal cycle is a crucial factor in the life of organisms, allowing for species succession and interaction. This study investigates the impact of changes in trophic status and thermal regime on a warm-adapted copepod species in Lake Biwa. The findings reveal that the copepod exhibited different seasonal cycles during the study period, with disruptions in the common unimodal cycle. The study highlights the importance of considering the seasonality of both lower and higher trophic levels in understanding zooplankton phenology.
JOURNAL OF GREAT LAKES RESEARCH
(2022)
Article
Biochemistry & Molecular Biology
Mark Louie D. Lopez, Ya-ying Lin, Stephan Q. Schneider, Chih-hao Hsieh, Fuh-Kwo Shiah, Ryuji J. Machida
Summary: The study extended assumptions of growth-rate hypothesis and metabolic theory of ecology to validate the allometric scaling of interspecific RNA transcript abundance. Results showed that body size imposes significant constraints on RNA transcript abundance, with larger individuals having lesser RNA transcript abundance per tissue mass than smaller ones. The relationship between total mitochondrial transcript abundance and body size aligned with metabolic theory assumptions, while nuclear transcripts displayed steeper slopes.
MOLECULAR ECOLOGY RESOURCES
(2023)
Correction
Multidisciplinary Sciences
Chun-Wei Chang, Takeshi Miki, Hao Ye, Sami Souissi, Rita Adrian, Orlane Anneville, Helen Agasild, Syuhei Ban, Yaron Be'eri-Shlevin, Yin-Ru Chiang, Heidrun Feuchtmayr, Gideon Gal, Satoshi Ichise, Maiko Kagami, Michio Kumagai, Xin Liu, Shin-Ichiro S. Matsuzaki, Marina M. Manca, Peeter Noges, Roberta Piscia, Michela Rogora, Fuh-Kwo Shiah, Stephen J. Thackeray, Claire E. Widdicombe, Jiunn-Tzong Wu, Tamar Zohary, Chih-hao Hsieh
NATURE COMMUNICATIONS
(2022)
Article
Multidisciplinary Sciences
Florian Grziwotz, Chun -Wei Chang, Vasilis Dakos, Egbert H. van Nes, Markus Schwarzlaender, Oliver Kamps, Martin Hessler, Isao T. Tokuda, Arndt Telschow, Chih-hao Hsieh
Summary: Critical transitions occur in various real-world systems and forecasting their occurrence is of great interest. This study introduces a powerful early warning signal called dynamical eigenvalue (DEV) that estimates the dominant eigenvalue of a system using bifurcation theory. The efficacy of the DEV approach is demonstrated in model systems with known bifurcation types and tested on various critical transitions in real-world systems.
Article
Microbiology
Jihye Yang, Sohyeon Yun, Woojun Park
Summary: The presence or absence of BlsA protein in Acinetobacter species genomes has raised curiosity about its role in regulating bacterial lifestyle under light. Genetic and transcriptomic analyses revealed the loss of BlsA in several multidrug-resistant strains and the light-mediated induction of blsA. Furthermore, the study identified a potential interaction partner of BlsA, BipA, and confirmed their direct interactions under specific conditions.
Article
Microbiology
Feng-Hsun Chang, Jinny Wu Yang, Ariana Chih-Hsien Liu, Hsiao-Pei Lu, Gwo-Ching Gong, Fuh-Kwo Shiah, Chih-hao Hsieh
Summary: The presence of more species in a community increases ecosystem functions via nonrandom processes like resource partitioning. Additionally, higher compositional difference between multiple communities also enhances their overall functions, especially when the difference is due to nonrandom assembly processes. In a study of bacterioplankton in the southern East China Sea, it was found that beta diversity positively affects the overall function of communities, with the effect being stronger when nonrandom processes select for phylogenetically dissimilar species. This research expands the biodiversity-ecosystem functioning framework to multiple sites and considers community assembly processes.
Editorial Material
Environmental Sciences
Feng-Hsun Chang, Yun-Chi Lin, Kuo-Ping Chiang, Ya-Han Nien, Chih-hao Hsieh, Wei-Jen Chen, Chin-Chang Hung, Toru Kobari, Hiroaki Saito, William Savidge
FRONTIERS IN MARINE SCIENCE
(2023)
Article
Microbiology
Jinny Wu Yang, Feng-Hsun Chang, Yi-Chun Yeh, An-Yi Tsai, Kuo-Ping Chiang, Fuh-Kwo Shiah, Gwo-Ching Gong, Chih-hao Hsieh
Summary: We studied the kill-the-winner hypothesis in marine bacterial communities and found evidence supporting the existence of competition-resistance trade-offs and their positive effect on bacterial diversity. Our research showed that competition-resistance trade-offs were stronger and more consistent when top-down control was caused by protists+viruses combined. These findings provide new insights into how natural bacterial communities are shaped by top-down controls and competition trade-offs.