Neural graph pattern machine. Jan 30, 2025 · Graph learning tasks require models to comprehend essential substructure patterns relevant to downstream tasks, such as triadic closures in social networks and benzene rings in molecular graphs. To overcome this limitation, we propose the Neural Graph Pattern Machine (GPM), a framework designed to learn directly from graph patterns. . Despite their empirical success Abstract Graph learning tasks require models to compre-hend essential substructure patterns relevant to downstream tasks, such as triadic closures in social networks and benzene rings in molecu-lar graphs. Jan 30, 2025 · In this paper, we introduce the Neural Graph Pattern Machine (GPM), a novel framework that bypasses message passing by learning directly from graph substructures. GPM ef- ficiently extracts, encodes, and prioritizes task- relevant graph patterns, offering greater expres- sivity and improved ability to capture long-range dependencies. This is the official implementation of our ICML'25 paper Beyond Message Passing: Neural Graph Pattern Machine. Recently, I have focused on developing foundational, resource-efficient, and safe machine learning algorithms and models, particularly on graph and text data. Feb 2, 2025 · The Neural Graph Pattern Machine represents a significant advancement in graph processing and pattern recognition. In this paper, we introduce the Neural G raph P attern M achine (GPM), a novel framework that bypasses message passing by learning directly from graph substructures. Despite their Poster Beyond Message Passing: Neural Graph Pattern Machine Zehong Wang · Zheyuan Zhang · Tianyi MA · Nitesh Chawla · Chuxu Zhang · Yanfang Ye East Exhibition Hall A-B #E-3105 May 1, 2025 · In this paper, we introduce the Neural Graph Pattern Machine (GPM), a novel framework that bypasses message passing by learning directly from graph substructures. In this paper, we introduce the Neural Graph Pattern Machine (GPM), a novel frame-work that bypasses message passing by learn-ing directly from graph substructures. Due to the non-Euclidean nature of graphs, existing graph neural networks (GNNs) rely on message passing to iteratively aggregate information from local neighborhoods. My research focuses on machine learning, with a particular emphasis on developing effective, efficient, and generalizable models. GPM efficiently extracts, encodes, and prioritizes task-relevant graph patterns, offering greater expressivity and improved ability to capture long-range dependencies. Currently, I am deeply engaged in designing next generation graph learning models, exploring their methodologies, theoretical basis, and applications. Dec 31, 2024 · To overcome this limitation, we propose the Neural Graph Pattern Machine (GPM), a framework designed to learn directly from graph patterns. GPM represents a significant step towards next-generation graph learning backbone by moving beyond traditional message passing approaches. In this paper, we introduce the Neural Graph Pattern Machine (GPM), a novel frame- work that bypasses message passing by learn- ing directly from graph substructures. GPM ef-ficiently extracts, encodes, and prioritizes task-relevant graph patterns, offering greater expres-sivity and improved ability to capture long-range dependencies. GPM efficiently extracts and encodes substructures while identifying the most relevant ones for downstream tasks. My research interests center around machine learning and data science. Its efficient approach to handling complex graph structures opens new possibilities for applications in network analysis, social media monitoring, and molecular structure analysis. pcj ojkh no dsqu0 use dtsdu xmzmmf qpl buo qyybhp7