<ul data-eligibleForWebStory="false"><li>Circuit link prediction is important in analog circuit design automation, but current methods face challenges like insufficient use of circuit graph patterns, data scarcity, and limited adaptability to netlist formats.</li><li>GNN-ACLP is a Graph Neural Networks (GNNs) based framework that addresses these challenges with innovations like the SEAL framework for improved prediction accuracy, Netlist Babel Fish for format conversion, and SpiceNetlist dataset for model training.</li><li>Experiments show significant accuracy improvements on various datasets with GNN-ACLP, demonstrating robust feature transfer capabilities and enhanced model generalization.</li></ul>

GNN-ACLP: Graph Neural Networks Based Analog Circuit Link Prediction

Discover more