Article
Computer Science, Artificial Intelligence
Sirichanya Chanmee, Kraisak Kesorn
Summary: This study introduces a new approach called the Semantic Decision Tree (SDT) to effectively address the multi-value bias selection issue and improve the generation of decision tree nodes. Evaluation results on multiple datasets show that SDT outperforms traditional algorithms in terms of accuracy and aligns more naturally with human decision-making logic.
ADVANCED ENGINEERING INFORMATICS
(2023)
Article
Engineering, Environmental
Marc Rovira, Klas Engvall, Christophe Duwig
Summary: This study examines the capabilities of a data-driven workflow for automated key feature identification in reactive flows. The proposed workflow aims to accelerate the analysis of chemical engineering datasets by generating automatic and explainable classification results for regions with distinct physics. The three main steps of the workflow, namely dimensionality reduction, unsupervised clustering, and feature correlation, are discussed. The study demonstrates the theoretical and practical differences between the previous and current algorithms used in the workflow. The updated workflow is shown to have faster, more accurate, and more robust key feature identification capabilities, closer to human intuition than previous methods. The study also serves as a tutorial for researchers interested in applying these algorithms.
CHEMICAL ENGINEERING JOURNAL
(2022)
Article
Computer Science, Artificial Intelligence
Hongyuan Zhang, Yanan Zhu, Xuelong Li
Summary: This study proposes a novel projected clustering framework to capture the essence of deep clustering by summarizing the core properties of powerful models, especially deep models. The framework introduces an aggregated mapping, consisting of projection learning and neighbor estimation, to obtain clustering-friendly representation. The study also addresses the problem of severe degeneration in simple clustering-friendly representation learning, and develops a self-evolution mechanism to alleviate the risk of over-fitting.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2023)
Article
Mathematics, Interdisciplinary Applications
Antonio M. Lopes
Summary: This paper proposes an unsupervised machine learning technique to analyze global terrorism dynamics, identifying phases and phase transitions. The study uses a dataset of worldwide terrorist incidents from 1970 to 2019 to generate multidimensional time-series representing casualties and events. The time-series are sliced and the resulting segments are characterized as objects that capture the system dynamics. These objects are compared and categorized using multidimensional scaling (MDS), generating portraits that illustrate the patterns and nature of the dynamics. The results demonstrate the effectiveness of MDS in analyzing global terrorism and its potential for studying other complex systems.
INTERNATIONAL JOURNAL OF BIFURCATION AND CHAOS
(2023)
Article
Mathematics, Interdisciplinary Applications
Antonio M. Lopes, J. A. Tenreiro Machado
Summary: This paper proposes an approach based on unsupervised machine learning to identify phases and phase transitions in complex systems. By generating multidimensional time-series and analyzing them using multidimensional scaling technique, the study finds that this method is relevant for modeling the behavior of complex systems.
CHAOS SOLITONS & FRACTALS
(2022)
Article
Computer Science, Artificial Intelligence
Ali Hassani, Amir Iranmanesh, Najme Mansouri
Summary: This study introduces a new feature agglomeration method based on nonnegative matrix factorization and proposes a deterministic initialization method for spherical K-means algorithm, which significantly improves the stability and performance of text data clustering.
NEURAL COMPUTING & APPLICATIONS
(2021)
Article
Chemistry, Multidisciplinary
Umar Subhan Malhi, Junfeng Zhou, Cairong Yan, Abdur Rasool, Shahbaz Siddeeq, Ming Du
Summary: This paper proposes a fashion image clustering method based on deep clustering, which uses convolutional neural networks to generate high-dimensional feature vectors and then reduces dimensions through auto-encoders before performing clustering. By jointly learning and optimizing the dimensionality reduction process and the clustering task, the proposed method achieves state-of-the-art performance.
APPLIED SCIENCES-BASEL
(2023)
Article
Green & Sustainable Science & Technology
Remy Rigo-Mariani
Summary: The paper proposes a strategy for reducing time horizon in power and energy studies. The proposed method displays smaller errors, is more scalable, and has less impact on system operation compared to conventional approaches.
RENEWABLE & SUSTAINABLE ENERGY REVIEWS
(2022)
Article
Computer Science, Artificial Intelligence
Zheng Liu, Wei Jin, Ying Mu
Summary: This paper introduces a unified framework that incorporates robust graph learning and dimensionality reduction, as well as clustering task. Two robust graph methods based on Euclidean distance and self-expressiveness are proposed, which are informative, robust, and sparse. Extensive experiments demonstrate their advantages in the task of clustering.
INTERNATIONAL JOURNAL OF INTELLIGENT SYSTEMS
(2022)
Article
Environmental Sciences
Divyesh Varade, Ajay K. Maurya, Onkar Dikshit
Summary: Information on snow cover distribution is important in hydrological processes and climate models. Hyperspectral remote sensing provides opportunities in land cover assessment, but is limited in snow-covered alpine regions due to large dimensionality. A band selection technique based on mutual information is proposed to improve efficiency and accuracy in selecting informative bands.
GEOCARTO INTERNATIONAL
(2021)
Article
Computer Science, Artificial Intelligence
Erick Odhiambo Omuya, George Onyango Okeyo, Michael Waema Kimwele
Summary: This study investigates the application of feature selection and classification in various fields, addressing the challenges of high dimensionality in datasets and the negative impact of irrelevant and redundant attributes on classification algorithms. To improve classification performance, a hybrid filter model based on principal component analysis and information gain is proposed and applied to machine learning techniques, demonstrating enhanced accuracy, precision, and recall.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Computer Science, Software Engineering
Mislene da Silva Nunes, Gastao Florencio Miranda Junior, Beatriz Trinchao Andrade
Summary: The search for realism in renderings has led to an increased interest in tabular BRDF samples captured from real-world materials. This study proposes an approach to generate new BRDFs based on user-selected materials from a database, creating an appearance-driven space using dimensionality reduction and clustering techniques.
COMPUTERS & GRAPHICS-UK
(2022)
Article
Automation & Control Systems
Mayank Jain, Tarek AlSkaif, Soumyabrata Dev
Summary: This article introduces a novel scheme to objectively validate and compare the clustering results of residential electric demand profiles, considering all steps prior to the clustering algorithm. Compared to traditional clustering validity indices, the proposed scheme provides better, unbiased, and uniform recommendations.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2021)
Article
Agriculture, Multidisciplinary
Sergio Velez, Florian Rancon, Enrique Barajas, Guilhem Brunel, Jose Antonio Rubio, Bruno Tisseyre
Summary: This study utilizes Sentinel-2 satellite imagery to extract relevant information from two vineyards in Spain. By employing dimensionality reduction techniques such as Principal Component Analysis (PCA) and Partial Least Square (PLS), the NDVI time-series are decomposed into multiple functional components. The results demonstrate the added value of considering the entire time-series compared to a single image, and establish correlations with seasonal phenology and management practices in the vineyards.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2022)
Article
Engineering, Electrical & Electronic
Yuheng Jia, Hui Liu, Junhui Hou, Sam Kwong, Qingfu Zhang
Summary: This paper introduces a self-supervised symmetric nonnegative matrix factorization (SNMF) method to improve data clustering performance. By exploiting the sensitivity to initialization of SNMF, without relying on additional information, the method progressively enhances clustering results. Experimental results demonstrate its superiority over 14 state-of-the-art methods in terms of multiple quantitative metrics.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2022)
Article
Ecology
Patricia A. Soranno, Kendra Spence Cheruvelil, Boyang Liu, Qi Wang, Pang-Ning Tan, Jiayu Zhou, Katelyn B. S. King, Ian M. McCullough, Jemma Stachelek, Meridith Bartley, Christopher T. Filstrup, Ephraim M. Hanks, Jean-Francois Lapierre, Noah R. Lottig, Erin M. Schliep, Tyler Wagner, Katherine E. Webster
ECOLOGICAL APPLICATIONS
(2020)
Article
Health Care Sciences & Services
Bum Chul Kwon, Courtland VanDam, Stephanie E. Chiuve, Hyung Wook Choi, Paul Entler, Pang-Ning Tan, Jina Huh-Yoo
JMIR MHEALTH AND UHEALTH
(2020)
Article
Computer Science, Artificial Intelligence
Jianpeng Xu, Jiayu Zhou, Pang-Ning Tan, Xi Liu, Lifeng Luo
Summary: Predictive modeling of large-scale spatio-temporal data is a challenging problem that requires training models to predict target variables at multiple locations while preserving spatial and temporal dependencies. This paper explores the effectiveness of using supervised tensor decomposition for multi-task learning in spatio-temporal prediction. The proposed framework, SMART, encodes data as a third-order tensor and trains ensemble models based on interpretable latent factors extracted from the data to make predictions on test instances, incorporating known patterns as constraints.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2021)
Article
Public, Environmental & Occupational Health
Young Anna Argyris, Kafui Monu, Pang-Ning Tan, Colton Aarts, Fan Jiang, Kaleigh Anne Wiseley
Summary: The study compared the discursive topics chosen by pro- and antivaccine advocates in influencing the public, finding that antivaccine topics have greater intertopic distinctiveness and use all four message frames, while provaccine advocates have neglected having a clear problem statement.
JMIR PUBLIC HEALTH AND SURVEILLANCE
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Tyler Wilson, Pang-Ning Tan, Lifeng Luo
Summary: This paper presents a deep learning framework for long-term prediction of the distribution of extreme values at different locations and addresses the computational challenges associated with large-scale geospatio-temporal data.
THIRTY-SIXTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE / THIRTY-FOURTH CONFERENCE ON INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE / THE TWELVETH SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Boyang Liu, Pang-Ning Tan, Jiayu Zhou
Summary: This paper proposes a robust deep density estimation framework for unsupervised anomaly detection, which improves the performance by discarding data points with low estimated densities and applying Lipschitz regularization.
THIRTY-SIXTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE / THIRTY-FOURTH CONFERENCE ON INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE / THE TWELVETH SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Francisco Santos, Junke Ye, Farzan Masrour, Pang-Ning Tan, Abdol-Hossein Esfahanian
Summary: Graph neural networks (GNNs) are widely used for modeling graph data by integrating node attributes and link information into concise representations. However, node classification using GNNs faces challenges such as imbalanced class distribution and the bias caused by the homophily effect. To address these challenges, we propose a novel framework called Fairness-Aware Cost Sensitive Graph Convolutional Network (FACS-GCN) that combines a cost-sensitive exponential loss and adversarial learning to achieve fair classification.
2022 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN)
(2022)
Proceedings Paper
Computer Science, Interdisciplinary Applications
Young Anna Argyris, Nan Zhang, Bidhan Bashyal, Pang-Ning Tan
Summary: This study aims to investigate the linguistic features of vaccine-related content and their impact on propagation, identifying two sets of features that either facilitate or inhibit the spread of vaccine-related tweets. Results show that anti-vaccine tweets tend to be propagated through retweets, while pro-vaccine tweets mainly receive passive endorsements.
2022 IEEE INTERNATIONAL CONFERENCE ON DIGITAL HEALTH (IEEE ICDH 2022)
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Ding Wang, Pang-Ning Tan
Summary: This paper introduces a novel online learning framework called JOHAN, which simultaneously predicts the trajectory and intensity of hurricanes, generates accurate forecasts of hurricane intensity categories, and uses exponentially-weighted quantile loss functions to improve prediction accuracy for high category hurricanes approaching landfall. Experimental results show the superiority of JOHAN over several state-of-the-art learning approaches using real-world hurricane data.
KDD '21: PROCEEDINGS OF THE 27TH ACM SIGKDD CONFERENCE ON KNOWLEDGE DISCOVERY & DATA MINING
(2021)
Article
Computer Science, Information Systems
Pouyan Hatami Bahman Beiglou, Lifeng Luo, Pang-Ning Tan, Lisi Pei
Summary: The US Drought Monitor is a vital tool for real-time drought monitoring, but its production involves human judgment, making it difficult for others to reproduce the maps. This study developed a framework using machine learning to automatically generate similar maps, with the support vector machines algorithm and specific data group achieving near-perfect reproduction accuracy.
FRONTIERS IN BIG DATA
(2021)
Article
Communication
Saleem Alhabash, Duygu Kanver, Chen Lou, Sandi W. Smith, Pang-Ning Tan
Summary: The study found that underage youth's perception of societal and personal celebration drinking norms were related to their close friends' drinking norms, which influenced their alcohol consumption during Halloween. Additionally, social media posting and interaction with alcohol-related content were associated with greater descriptive normative perceptions and self-reported drinking.
HEALTH COMMUNICATION
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Farzan Masrour, Pang-Ning Tan, Abdol-Hossein Esfahanian
20TH IEEE INTERNATIONAL CONFERENCE ON DATA MINING (ICDM 2020)
(2020)
Proceedings Paper
Computer Science, Artificial Intelligence
Tyler Wilson, Pang-Ning Tan, Lifeng Luo
PROCEEDINGS OF THE 2020 SIAM INTERNATIONAL CONFERENCE ON DATA MINING (SDM)
(2020)
Article
Limnology
Tyler Wagner, Noah R. Lottig, Meridith L. Bartley, Ephraim M. Hanks, Erin M. Schliep, Nathan B. Wikle, Katelyn B. S. King, Ian McCullough, Jemma Stachelek, Kendra S. Cheruvelil, Christopher T. Filstrup, Jean Francois Lapierre, Boyang Liu, Patricia A. Soranno, Pang-Ning Tan, Qi Wang, Katherine Webster, Jiayu Zhou
LIMNOLOGY AND OCEANOGRAPHY LETTERS
(2020)
Article
Communication
Sandi W. Smith, Saleem Alhabash, Duygu Kanver, Pang-Ning Tan, Greg Viken
HEALTH COMMUNICATION
(2020)