Article
Energy & Fuels
Yi Wang, Luzian Lebovitz, Kedi Zheng, Yao Zhou
Summary: A novel weighted consensus clustering-based approach is proposed for bi-objective power network partition, allowing for Pareto improvement. Case studies on the IEEE 300-bus test system demonstrate the effectiveness and superiority of the proposed method.
CSEE JOURNAL OF POWER AND ENERGY SYSTEMS
(2022)
Article
Chemistry, Medicinal
Alejandro Gomez-Garcia, Daniel A. Acuna Jimenez, William J. Zamora, Haruna L. Barazorda-Ccahuana, Miguel A. Chavez-Fumagalli, Marilia Valli, Adriano D. Andricopulo, Vanderlan da S. Bolzani, Dionisio A. Olmedo, Pablo N. Solis, Marvin J. Nunez, Johny R. Rodriguez Perez, Hoover A. Valencia Sanchez, Hector F. Cortes Hernandez, Jose L. Medina-Franco
Summary: The Latin American Natural Products Database (LANaPDB) is a public compound collection that gathers chemical information of natural products from various databases in Latin America. The current version includes chemical structures from six countries, with terpenoids, phenylpropanoids, and alkaloids being the most abundant compounds. LANaPDB covers a wide chemical space and many compounds exhibit drug-like properties.
Review
Pharmacology & Pharmacy
Fernanda Saldivar-Gonzalez, Jose L. Medina-Franco
Summary: This article reviews the current state of chemical space in drug design and discovery, discussing advances in efficient navigation and assessment of chemical space diversity, as well as highlighting recent methods for generating visual representations of chemical space.
EXPERT OPINION ON DRUG DISCOVERY
(2022)
Article
Computer Science, Artificial Intelligence
Pei Zhang, Xinwang Liu, Jian Xiong, Sihang Zhou, Wentao Zhao, En Zhu, Zhiping Cai
Summary: This paper proposes a novel consensus one-step multi-view subspace clustering (COMVSC) method, which eliminates noise by optimally integrating partition-level information and learns affinity matrices, consensus representation, and final clustering labels matrix simultaneously in a unified framework. Experimental results demonstrate the superiority of our method against other state-of-the-art approaches.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2022)
Article
Engineering, Electrical & Electronic
Jinhao Qian, Xiaohua Xie
Summary: This paper focuses on unsupervised video-based person re-identification and proposes a Successive Consensus Clustering framework for optimizing pseudo-labels and the model. By leveraging multiple frames clustering and a cluster successive memory mechanism, the method achieves improved model training and performance.
IEEE SIGNAL PROCESSING LETTERS
(2022)
Article
Biochemistry & Molecular Biology
Dipro Sinha, Anu Sharma, Dwijesh Chandra Mishra, Anil Rai, Shashi Bhushan Lal, Sanjeev Kumar, Moh. Samir Farooqi, Krishna Kumar Chaturvedi
Summary: This paper introduces a novel method, MetaConClust, using coverage information for binning of metagenomic data and automatically finding the optimal number of clusters. Experimental results show that this method performs better for unsupervised techniques and similar for hybrid methods.
Article
Computer Science, Artificial Intelligence
He Sun, Lei Zhang, Jinchang Ren, Hua Huang
Summary: To address the issue of inconsistent band selection in the dimensionality reduction of HSI, this study proposes a hyperbolic clustering-based band hierarchy method, which can better represent the data structure and achieve consistent band selection.
PATTERN RECOGNITION
(2022)
Article
Computer Science, Artificial Intelligence
Yujing Zhang, Siwei Wang, Xinwang Liu, En Zhu
Summary: In this study, a late fusion MKC method with local graph refinement is proposed to address the issues of representation flexibility and locality structure preservation in current MKC methods. By unifying traditional weighted multiple kernel k-means, kernel partition, and graph construction into a single optimization procedure, the proposed method enhances the local graph structure in partition space and allows for more flexible kernel representations, resulting in significant clustering improvements. Experiments on multiple kernel benchmark datasets demonstrate the effectiveness and superiority of the proposed algorithm compared to state-of-the-art methods.
INTERNATIONAL JOURNAL OF INTELLIGENT SYSTEMS
(2021)
Article
Computer Science, Artificial Intelligence
Xianfeng Ou, Meng Wu, Bing Tu, Guoyun Zhang, Wujing Li
Summary: With the increasing spectral dimension of hyperspectral images, the correct choice of bands based on band correlation and information has become more significant and complicated. To address this, we propose a band selection method based on a multi-objective cuckoo search algorithm, which constructs a multi-objective unsupervised band selection model. The proposed method outperforms state-of-the-art algorithms in HSI classification experiments, demonstrating its effectiveness and robustness.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2023)
Article
Mathematics
Xiuqin Deng, Yifei Zhang, Fangqing Gu
Summary: Multi-view subspace clustering is an effective method that has been successfully applied in various applications and has attracted attention from scholars. However, most existing methods only focus on multi-view information and ignore feature concatenation, leading to a failure in exploring their high correlation. To address this, this paper proposes a novel consensus matrix construction strategy for multi-view subspace clustering. The proposed algorithm learns a consensus matrix by fusing information from multiple views and enhances it with the original feature direct linkage. Experimental results on six datasets demonstrate the effectiveness of the proposed algorithm.
Article
Chemistry, Analytical
Najmeh Razfar, Rasha Kashef, Farah Mohammadi
Summary: Stroke survivors often experience movement impairments that affect their daily activities. Advances in sensor technology and IoT have opened up possibilities for automating post-stroke assessment and rehabilitation using AI-driven models. This paper introduces a smart post-stroke severity assessment approach that uses unsupervised learning and trunk displacement features in the frequency domain. The proposed consensus clustering algorithm, PSA-NMF, combines different clusterings to produce more stable and robust results. Experimental results show improved evaluation metrics, such as accuracy and F-score, which can contribute to a more effective and automated stroke rehabilitation process in clinical settings.
Article
Engineering, Civil
Aihua Zheng, Xia Sun, Chenglong Li, Jin Tang
Summary: Vehicle re-identification is crucial for large-scale intelligent monitoring in smart cities, but existing methods are mostly supervised, time-consuming, and limited in real-life scenarios. Unsupervised person re-identification methods show impressive performance through domain adaption or clustering techniques, but cannot be directly applied to vehicle re-identification due to huge appearance variations in vehicle images. To address this issue, a novel viewpoint-aware clustering algorithm is proposed for unsupervised vehicle re-identification, achieving promising performance in various scenarios according to experiments on benchmark datasets.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Computer Science, Software Engineering
Soumaya Rebai, Vahid Alizadeh, Marouane Kessentini, Houcem Fehri, Rick Kazman
Summary: This paper proposes an interactive approach that allows developers to pinpoint their preferences in both the objective and decision spaces, resulting in more efficient resolution of quality issues. By using multi-objective search and clustering algorithms, developers are able to examine a smaller number of solutions and provide feedback, which is then used to generate constraints and optimize the refactoring process.
IEEE TRANSACTIONS ON SOFTWARE ENGINEERING
(2022)
Article
Chemistry, Medicinal
Yuliana Zabolotna, Peter Ertl, Dragos Horvath, Fanny Bonachera, Gilles Marcou, Alexandre Varnek
Summary: NP Navigator is a freely available online tool based on compounds from COCONUT, ChEMBL and ZINC, allowing visualization and navigation through the chemical space of NPs and NP-like molecules, enabling efficient analysis of different aspects of NPs.
MOLECULAR INFORMATICS
(2021)
Article
Computer Science, Information Systems
Na Li, Arnaud Martin, Remi Estival
Summary: In real-life machine learning applications, combining supervised classification and clustering results at the output level is gaining attention to improve accuracy when raw data is inaccessible or training samples are limited. This approach helps reduce dependency on raw data and uncertainty in supervised results, while studying the impact of supervised classification and clustering results on output combination.
INFORMATION SCIENCES
(2021)
Article
Chemistry, Medicinal
Edgar Lopez-Lopez, Carlos M. Cerda-Garcia-Rojas, Jose L. Medina-Franco
Summary: Modification of the tubulin-microtubule system has led to effective strategies for cancer treatment. However, many inhibitors of this system are based on limited structural innovation. This study aims to develop a screening protocol for natural products to identify new potential inhibitors of the tubulin-microtubule system.
MOLECULAR INFORMATICS
(2023)
Review
Biochemistry & Molecular Biology
Alejandro Gomez-Garcia, Jose L. Medina-Franco
Summary: This article reviews the progress in developing compound databases of natural origin, with a focus on chemoinformatic approaches. It also surveys the present state of developing Latin American NP databases and their practical applications in drug discovery.
Article
Biochemistry & Molecular Biology
Barbara Diaz-Eufracio, Jose L. Medina-Franco
Summary: This study develops a classification model to identify protein-protein interaction inhibitors using machine learning algorithms and chemoinformatics techniques, and provides free code. The results show that different algorithms and molecular fingerprints have varying performances in the training process, with random forest models trained using extended connectivity fingerprint radius 2 having the best classification abilities.
Article
Chemistry, Medicinal
Timothy B. Dunn, Edgar Lopez-Lopez, Taewon David Kim, Jose L. Medina-Franco, Ramon Alain Miranda-Quintana
Summary: "Understanding structure-activity landscapes" emphasizes the importance of the relationship between molecular structure and activity in drug discovery, and highlights the impact of activity cliffs on design and prediction. This study aims to use n-ary indices and the medoid algorithm to rapidly and efficiently quantify the structure-activity landscapes of large compound data sets, and explore the optimal correlations between similarity measures and structure-activity rankings. The applicability of these methods is demonstrated by analyzing 10 compound data sets with pharmaceutical relevance using different fingerprints, similarity indices, and coincidence thresholds.
MOLECULAR INFORMATICS
(2023)
Article
Biochemistry & Molecular Biology
Kenneth Lopez-Perez, Edgar Lopez-Lopez, Jose L. Medina-Franco, Ramon Alain Miranda-Quintana
Summary: Visualization of chemical space is crucial in various aspects of chemistry, such as compound library design and exploring structure-property relationships. ChemMaps is a visualization methodology that approximates compound distribution in large datasets based on satellite compounds with a similar mapping. Extended similarity indices are proposed to identify relevant regions and reduce high-dimensional data in describing a library's chemical space.
Review
Biochemistry & Molecular Biology
Cesar A. Zaa, alvaro J. Marcelo, Zhiqiang An, Jose L. Medina-Franco, Marco A. Velasco-Velazquez
Summary: Anthocyanins, a type of flavonoid, possess potent antioxidant properties and can penetrate the blood-brain barrier, exhibiting neuroprotective effects. Including anthocyanin-rich foods in the diet has been shown to lower the risk of neurodegenerative diseases, as their antioxidant, anti-inflammatory, and anti-apoptotic properties contribute to neuroprotection, particularly in Alzheimer's and Parkinson's diseases.
Article
Chemistry, Multidisciplinary
Raziel Cedillo-Gonzalez, Jose L. Medina-Franco
Summary: This study explores, characterizes, and analyzes the chemical space of 409 G9a inhibitors reported in a large public database. It quantifies the structural diversity of G9a inhibitors and compares them with commercial libraries focused on epigenetic targets. The findings will contribute to the development of predictive models for identifying G9a inhibitors and highlight the importance of screening commercial libraries to expand the relevant chemical space.
Article
Chemistry, Medicinal
Alejandro Gomez-Garcia, Daniel A. Acuna Jimenez, William J. Zamora, Haruna L. Barazorda-Ccahuana, Miguel A. Chavez-Fumagalli, Marilia Valli, Adriano D. Andricopulo, Vanderlan da S. Bolzani, Dionisio A. Olmedo, Pablo N. Solis, Marvin J. Nunez, Johny R. Rodriguez Perez, Hoover A. Valencia Sanchez, Hector F. Cortes Hernandez, Jose L. Medina-Franco
Summary: The Latin American Natural Products Database (LANaPDB) is a public compound collection that gathers chemical information of natural products from various databases in Latin America. The current version includes chemical structures from six countries, with terpenoids, phenylpropanoids, and alkaloids being the most abundant compounds. LANaPDB covers a wide chemical space and many compounds exhibit drug-like properties.
Article
Chemistry, Multidisciplinary
Edgar Lopez-Lopez, Oscar Robles, Fabien Plisson, Jose L. Medina-Franco
Summary: Peptide structure-activity/property relationship (P-SA/PR) studies aim to understand how the structural variations of peptides influence their biological activities and other functional properties, accelerating the rational design and optimization of peptide-based drugs, biomaterials, or diagnostic agents. This study used the MAP4 fingerprint to analyze the structure-activity relationship of 223 antimicrobial peptides against methicillin-resistant Staphylococcus aureus (MRSA), identifying critical residues and structural motifs that play a crucial role in the anti-MRSA activity of the peptides.
Review
Chemistry, Multidisciplinary
Conrad V. Simoben, Smith B. Babiaka, Aurelien F. A. Moumbock, Cyril T. Namba-Nzanguim, Donatus Bekindaka Eni, Jose L. Medina-Franco, Stefan Guenther, Fidele Ntie-Kang, Wolfgang Sippl
Summary: The use of traditional medicine has a long history and is still relied upon by many, especially in developing or underprivileged communities. In silico-based methods have played a crucial role in drug discovery, particularly in identifying natural product-based candidates. However, there are challenges in identifying and proposing novel natural product-based hits.
