{"668989":{"#nid":"668989","#data":{"type":"event","title":"ISYE Statistic Seminar - Annie Qu","body":[{"value":"\u003Cp\u003EA Model-Agnostic Graph Neural Network for Integrating Local and Global Information\u003C\/p\u003E\r\n\r\n\u003Cp\u003EAbstract\u003C\/p\u003E\r\n\r\n\u003Cp\u003EGraph neural networks (GNNs) have achieved promising performance in a variety of graph focused\u003C\/p\u003E\r\n\r\n\u003Cp\u003Etasks. Despite their success, the two major limitations of existing GNNs are the capability\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eof learning various-order representations and providing interpretability of such deep learning-based\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eblack-box models. To tackle these issues, we propose a novel Model-agnostic Graph Neural\u003C\/p\u003E\r\n\r\n\u003Cp\u003ENetwork (MaGNet) framework. The proposed framework is able to extract knowledge from\u003C\/p\u003E\r\n\r\n\u003Cp\u003Ehigh-order neighbors, sequentially integrates information of various orders, and offers explanations\u003C\/p\u003E\r\n\r\n\u003Cp\u003Efor the learned model by identifying influential compact graph structures. In particular,\u0026nbsp;MaGNet\u003C\/p\u003E\r\n\r\n\u003Cp\u003Econsists of two components: an estimation model for the latent representation of complex\u003C\/p\u003E\r\n\r\n\u003Cp\u003Erelationships under graph topology, and an interpretation model that identifies influential nodes,\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eedges, and important node features. Theoretically, we establish the generalization error bound for\u003C\/p\u003E\r\n\r\n\u003Cp\u003EMaGNet\u0026nbsp;via empirical Rademacher complexity and showcase its power to represent the layer-wise\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eneighborhood mixing. We conduct comprehensive numerical studies using both simulated data\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eand a real-world case study on investigating the neural mechanisms of the rat hippocampus,\u003C\/p\u003E\r\n\r\n\u003Cp\u003Edemonstrating that the performance of\u0026nbsp;MaGNet\u0026nbsp;is competitive with state-of-the-art methods.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EBio:\u0026nbsp;\u003C\/p\u003E\r\n\r\n\u003Cp\u003EAnnie Qu\u003C\/p\u003E\r\n\r\n\u003Cp\u003EChancellor\u2019s Professor, Department of Statistics, University of California Irvine\u003C\/p\u003E\r\n\r\n\u003Cp\u003EPh.D., Statistics, the Pennsylvania State University\u003C\/p\u003E\r\n\r\n\u003Cp\u003EQu\u2019s research focuses on solving fundamental issues regarding structured and unstructured large-scale data, and developing cutting-edge statistical methods and theory in machine learning and algorithms on personalized medicine, text mining, recommender systems, medical imaging data and network data analyses for complex heterogeneous data. The newly developed methods are able to extract essential and relevant information from large volume high-dimensional data. Her research has impacts in many fields such as biomedical studies, genomic research, public health research, social and political sciences.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EBefore she joins the UC Irvine, Dr. Qu is Data Science Founder Professor of Statistics, and the Director of the Illinois Statistics Office at the University of Illinois at Urbana-Champaign. She was awarded as Brad and Karen Smith Professorial Scholar by the College of LAS at UIUC, a recipient of the NSF Career award in 2004-2009. She is a Fellow of the Institute of Mathematical Statistics, a Fellow of the American Statistical Association, and a Fellow of American Association for the Advancement of Science. She is also a recipient of Medallion Award and Lecturer. She is JASA Theory and Methods co-editor in 2023-2025.\u003C\/p\u003E\r\n","summary":"","format":"limited_html"}],"field_subtitle":"","field_summary":[{"value":"\u003Cp\u003EGraph neural networks (GNNs) have achieved promising performance in a variety of graph focused\u003C\/p\u003E\r\n\r\n\u003Cp\u003Etasks. Despite their success, the two major limitations of existing GNNs are the capability\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eof learning various-order representations and providing interpretability of such deep learning-based\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eblack-box models. To tackle these issues, we propose a novel Model-agnostic Graph Neural\u003C\/p\u003E\r\n\r\n\u003Cp\u003ENetwork (MaGNet) framework. The proposed framework is able to extract knowledge from\u003C\/p\u003E\r\n\r\n\u003Cp\u003Ehigh-order neighbors, sequentially integrates information of various orders, and offers explanations\u003C\/p\u003E\r\n\r\n\u003Cp\u003Efor the learned model by identifying influential compact graph structures. In particular,\u0026nbsp;MaGNet\u003C\/p\u003E\r\n\r\n\u003Cp\u003Econsists of two components: an estimation model for the latent representation of complex\u003C\/p\u003E\r\n\r\n\u003Cp\u003Erelationships under graph topology, and an interpretation model that identifies influential nodes,\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eedges, and important node features. Theoretically, we establish the generalization error bound for\u003C\/p\u003E\r\n\r\n\u003Cp\u003EMaGNet\u0026nbsp;via empirical Rademacher complexity and showcase its power to represent the layer-wise\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eneighborhood mixing. We conduct comprehensive numerical studies using both simulated data\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eand a real-world case study on investigating the neural mechanisms of the rat hippocampus,\u003C\/p\u003E\r\n\r\n\u003Cp\u003Edemonstrating that the performance of\u0026nbsp;MaGNet\u0026nbsp;is competitive with state-of-the-art methods.\u003C\/p\u003E\r\n","format":"limited_html"}],"field_summary_sentence":[{"value":"A Model-Agnostic Graph Neural Network for Integrating Local and Global Information"}],"uid":"36433","created_gmt":"2023-08-16 13:06:16","changed_gmt":"2023-08-16 13:06:16","author":"mrussell89","boilerplate_text":"","field_publication":"","field_article_url":"","field_event_time":{"event_time_start":"2023-09-12T11:00:00-04:00","event_time_end":"2023-09-12T12:00:00-04:00","event_time_end_last":"2023-09-12T12:00:00-04:00","gmt_time_start":"2023-09-12 15:00:00","gmt_time_end":"2023-09-12 16:00:00","gmt_time_end_last":"2023-09-12 16:00:00","rrule":null,"timezone":"America\/New_York"},"location":"Groseclose 402","extras":[],"groups":[{"id":"1242","name":"School of Industrial and Systems Engineering (ISYE)"}],"categories":[],"keywords":[],"core_research_areas":[],"news_room_topics":[],"event_categories":[{"id":"1795","name":"Seminar\/Lecture\/Colloquium"}],"invited_audience":[{"id":"78761","name":"Faculty\/Staff"},{"id":"177814","name":"Postdoc"},{"id":"78771","name":"Public"},{"id":"174045","name":"Graduate students"},{"id":"78751","name":"Undergraduate students"}],"affiliations":[],"classification":[],"areas_of_expertise":[],"news_and_recent_appearances":[],"phone":[],"contact":[],"email":[],"slides":[],"orientation":[],"userdata":""}}}