Article
Computer Science, Artificial Intelligence
Asmaa Rassil, Hiba Chougrad, Hamid Zouaki
Summary: Graph Neural Networks (GNNs) are powerful architectures for learning and processing graph data. This paper proposes the Augmented Graph Neural Network (AGNN) model and the Reversible Augmented Graph Neural Network (R-AGNN) model, which improve the accuracy of graph property prediction tasks through hierarchical residual connections and reversible mechanisms respectively, achieving state-of-the-art results on datasets from various domains.
Article
Computer Science, Artificial Intelligence
Asmaa Rassil, Hiba Chougrad, Hamid Zouaki
Summary: This paper proposes a Holistic Graph Neural Network (HGNN) that introduces a global-based attention mechanism to learn and generate node embeddings for improved performance on graph-structured data. Experimental results demonstrate the significant benefits of this approach compared to state-of-the-art methods.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Xuefeng Zhang, Richong Zhang, Junfan Chen, Jaein Kim, Yongyi Mao
Summary: This study proposes a model called GALA to address the limitation of existing entity alignment models in exploring the topology information of unaligned entities. By constructing global features and aggregating local information, GALA aligns entities from different knowledge graphs and achieves promising results.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2023)
Article
Environmental Sciences
Yukun Yang, Bo Ma, Xiangdong Liu, Liang Zhao, Shoudong Huang
Summary: The paper proposes a new method for visual place recognition, which extracts information from RGB and depth images and fuses them in graph data, treating the recognition problem as a graph classification issue. By using the Global Structure Attention Pooling method, the classification accuracy is improved.
Article
Computer Science, Information Systems
Pengbo Li, Hang Yu, Xiangfeng Luo, Jia Wu
Summary: Graphs are widely used in fraud detection tasks to capture complex features in scenarios with various relation attributes like transactions. However, existing methods mostly ignore global information, which is important in detecting local abnormal points. In this article, we propose a local and global aware memory-based graph neural network (LGM-GNN) that utilizes both local and global information through relation-aware embedding and interactive aggregation. Experimental results show that LGM-GNN outperforms other methods on real-world fraud detection datasets.
IEEE TRANSACTIONS ON BIG DATA
(2023)
Article
Engineering, Electrical & Electronic
Ji Zhang, Jean-Paul Ainam, Wenai Song, Li-hui Zhao, Xin Wang, Hongzhou Li
Summary: This paper proposes an end-to-end deep learning framework to overcome the limitations in person re-identification task by combining global and local feature representations and capturing body structural information using graph neural networks.
SIGNAL PROCESSING-IMAGE COMMUNICATION
(2022)
Article
Computer Science, Information Systems
Miaomiao Liu, Jingfeng Guo, Jing Chen, Yongsheng Zhang
Summary: A novel model was proposed for simultaneous link and sign prediction, with high accuracy for negative link prediction. Experiments showed the model's effectiveness on various datasets, achieving high prediction accuracy levels.
HUMAN-CENTRIC COMPUTING AND INFORMATION SCIENCES
(2021)
Article
Biochemical Research Methods
Sridevi Maharaj, Taotao Qian, Zarin Ohiba, Wayne Hayes
Summary: The joint distribution of degree products and common neighbors has a greater impact on PPI edge connectivity than their individual distributions, leading to the introduction of two new models (CN and STICKY-CN). The inclusion of CN into STICKY-CN makes it the best overall fit for PPI networks as it is a good fit locally and globally.
IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS
(2021)
Article
Computer Science, Artificial Intelligence
Darnbi Sakong, Thanh Trung Huynh, Thanh Tam Nguyen, Thanh Toan Nguyen, Jun Jo, Quoc Viet Hung Nguyen
Summary: This paper proposes a novel KG alignment framework, ComplexGCN, which learns the embeddings of both entities and relations in complex spaces while capturing both semantic and neighborhood information simultaneously. The proposed model ensures richer expressiveness and more accurate embeddings by successfully capturing various relation structures in complex spaces with high-level computation. The model further incorporates relation label and direction information with a low degree of freedom. Empirical results show the efficiency and effectiveness of the proposed method.
INTERNATIONAL JOURNAL OF INTELLIGENT SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Rui Wang, Yongkun Li, Shuai Lin, WeiJie Wu, Hong Xie, Yinlong Xu, John C. S. Lui
Summary: Random walk is a popular method for sampling large-scale graphs, but it suffers from slow convergence. To address this issue, we propose a common neighbor aware random walk framework called CNARW, which speeds up convergence by considering common neighbors. We also design efficient unbiased sampling schemes and variant algorithms to reduce cost and improve convergence. Experimental results show that our approach outperforms state-of-the-art random walk sampling algorithms in terms of convergence speed and query cost reduction. Two case studies demonstrate the effectiveness of our sampling framework in solving large-scale graph analysis tasks.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2023)
Article
Computer Science, Information Systems
Xiaoming Zhang, Wencheng Zhang, Huiyong Wang
Summary: Graph convolutional network-based methods are widely used for cross-language entity alignment. This study proposed a neighboring-entity-screening rule combining entity name and attribute to improve alignment results. Experimental results showed that the proposed method significantly improved overall entity alignment.
Article
Biochemical Research Methods
Hyun-Myung Woo, Byung-Jun Yoon
Summary: MONACO is a novel and versatile network alignment algorithm that achieves highly accurate pairwise and multiple network alignments through iterative optimal matching of 'local' neighborhoods around focal nodes. Extensive performance assessments on real and synthetic networks, where the ground truth is known, show that MONACO consistently outperforms all other state-of-the-art network alignment algorithms in terms of accuracy, coherence, and topological quality of the aligned network regions. Despite the significantly enhanced alignment accuracy, MONACO remains computationally efficient and scales well with increasing size and number of networks.
Article
Mathematical & Computational Biology
Jing Chen, Ying Zhang, Jin-Fang Xia
Summary: The development of high-throughput technology has provided a reliable technical guarantee for an increased amount of available data on biological networks. Network alignment is used to analyze these data to identify conserved functional network modules and understand evolutionary relationships across species. Hence, an efficient computational network aligner is needed for network alignment.
COMPUTATIONAL AND MATHEMATICAL METHODS IN MEDICINE
(2021)
Article
Physics, Multidisciplinary
Yuanhao Huang, Pengcheng Zhao, Qi Zhang, Ling Xing, Honghai Wu, Huahong Ma
Summary: User alignment is important for associating multiple social network accounts of the same user. However, the varying behaviors and friends of the same user across different social networks affect the accuracy of alignment. To mitigate this, a semantically enhanced social network user alignment algorithm (SENUA) is proposed. SENUA aligns users based on attributes, user-generated contents, and check-ins, while reducing local semantic noise by mining semantic features. The algorithm's adaptability to noise is improved using multi-view graph-data augmentation. Furthermore, embedding vectors are optimized using multi-headed graph attention networks and multi-view contrastive learning to enhance aligned users' similar semantic features. Experimental results demonstrate a 6.27% average improvement in hit-precision30, indicating the effectiveness of semantic enhancement in user alignment.
Article
Physics, Multidisciplinary
Herman Yuliansyah, Zulaiha Ali Othman, Azuraliza Abu Bakar
Summary: The cold-start problem occurs when a new user with limited information joins the network, making it challenging to predict new links in future networks. This study proposes a link prediction method, DGLP, enhanced by the gravity of node pairs inspired by Newton's law of gravity, to address the common neighbor's failure in predicting future relations for new users with cold-start problems.
PHYSICA A-STATISTICAL MECHANICS AND ITS APPLICATIONS
(2023)
Article
Biochemical Research Methods
Vladimir Gligorijevic, Noel Malod-Dognin, Natasa Przulj
Review
Biochemical Research Methods
Vladimir Gligorijevic, Noel Malod-Dognin, Natasa Przulj
Editorial Material
Multidisciplinary Sciences
Natasa Przulj, Noel Malod-Dognin
Article
Multidisciplinary Sciences
Michael Costanzo, Benjamin VanderSluis, Elizabeth N. Koch, Anastasia Baryshnikova, Carles Pons, Guihong Tan, Wen Wang, Matej Usaj, Julia Hanchard, Susan D. Lee, Vicent Pelechano, Erin B. Styles, Maximilian Billmann, Jolanda van Leeuwen, Nydia van Dyk, Zhen-Yuan Lin, Elena Kuzmin, Justin Nelson, Jeff S. Piotrowski, Tharan Srikumar, Sondra Bahr, Yiqun Chen, Raamesh Deshpande, Christoph F. Kurat, Sheena C. Li, Zhijian Li, Mojca Mattiazzi Usaj, Hiroki Okada, Natasha Pascoe, Bryan-Joseph San Luis, Sara Sharifpoor, Emira Shuteriqi, Scott W. Simpkins, Jamie Snider, Harsha Garadi Suresh, Yizhao Tan, Hongwei Zhu, Noel Malod-Dognin, Vuk Janjic, Natasa Przulj, Olga G. Troyanskaya, Igor Stagljar, Tian Xia, Yoshikazu Ohya, Anne-Claude Gingras, Brian Raught, Michael Boutros, Lars M. Steinmetz, Claire L. Moore, Adam P. Rosebrock, Amy A. Caudy, Chad L. Myers, Brenda Andrews, Charles Boone
Article
Biochemistry & Molecular Biology
Kate Sokolina, Saranya Kittanakom, Jamie Snider, Max Kotlyar, Pascal Maurice, Jorge Gandia, Abla Benleulmi-Chaachoua, Kenjiro Tadagaki, Atsuro Oishi, Victoria Wong, Ramy H. Malty, Viktor Deineko, Hiroyuki Aoki, Shahreen Amin, Zhong Yao, Xavier Morato, David Otasek, Hiroyuki Kobayashi, Javier Menendez, Daniel Auerbach, Stephane Angers, Natasa Przulj, Michel Bouvier, Mohan Babu, Francisco Ciruela, Ralf Jockers, Igor Jurisica, Igor Stagljar
MOLECULAR SYSTEMS BIOLOGY
(2017)
Article
Multidisciplinary Sciences
Anida Sarajlic, Noel Malod-Dognin, Omer Nebil Yaveroglu, Natasa Przulj
SCIENTIFIC REPORTS
(2016)
Article
Multidisciplinary Sciences
Noel Malod-Dognin, Kristina Ban, Natasa Przulj
SCIENTIFIC REPORTS
(2017)
Review
Pharmacology & Pharmacy
Sandra Kraljevic Pavelic, Jasmina Simovi Medica, Darko Gumbarevic, Ana Filosevic, Natasa Przulj, Kresimir Pavelic
FRONTIERS IN PHARMACOLOGY
(2018)
Article
Multidisciplinary Sciences
Noel Malod-Dognin, Julia Petschnigg, Sam F. L. Windels, Janez Povh, Harry Hemmingway, Robin Ketteler, Natasa Przulj
NATURE COMMUNICATIONS
(2019)
Article
Biochemical Research Methods
N. Malod-Dognin, V Pancaldi, A. Valencia, N. Przulj
Article
Biochemical Research Methods
D. A. Salazar, N. Przulj, C. F. Valencia
Summary: This study introduces a novel method capable of integrating multi-omic data from different sources and identifying patient and cell line groups enriched in cancer-associated gene clusters. The method also predicts drug profiles for patients and identifies genetic signatures for tumors resistant and sensitive to specific drugs.
Article
Biochemistry & Molecular Biology
Alexandros Xenos, Noel Malod-Dognin, Carme Zambrana, Natasa Przulj
Summary: To understand COVID-19, researchers used an integrated cell (iCell) concept with three omics networks to study its molecular basis. They compared patient-based and cell line-based iCells and found significant differences, indicating the limitations of using cell lines in studying this disease. By comparing infected and control patient-based iCells, they identified genes whose functioning is altered by the disease. They also predicted drugs for repurposing and confirmed the binding of these drugs to their targets. The study highlights the applicability of the iCell framework for studying human diseases.
INTERNATIONAL JOURNAL OF MOLECULAR SCIENCES
(2023)
Article
Mathematical & Computational Biology
Noel Malod-Dognin, Natasa Przulj
JOURNAL OF INTEGRATIVE BIOINFORMATICS
(2017)
Article
Biochemical Research Methods
Omer Nebil Yaveroglu, Noel Malod-Dognin, Tijana Milenkovic, Natasa Przulj
Proceedings Paper
Computer Science, Interdisciplinary Applications
Vladimir Gligorijevic, Noel Malod-Dognin, Natasa Przulj
PACIFIC SYMPOSIUM ON BIOCOMPUTING 2016
(2016)