2020 · 102 citations · 7 references
ComputingComputer ScienceElectrical EngineeringDesign AutomationDesign OptimizationMachine LearningDevice ModelingGraph AlgorithmsGraph-based LearningNetwork ModelingNeural NetworksVery-large-scale Integration LayoutCircuit Graph TheoryCircuit ModelingGraph Attention NetworksGraph Neural NetworksGraph-based PredictionGraph-theoretic Circuit FormulationLayout Parasitics ModelingNeural Network Modeling
Layout-dependent parasitics and device parameters significantly impact integrated circuit performance and are often the cause of slow convergences between schematic and layout designs. Circuit designers typically estimate parasitics from past experience, resulting in variability between designers and the potential for inaccuracies. In this paper, we present ParaGraph: a graph neural network model to predict net parasitics and device parameters by converting circuit schematics into graphs and leveraging key modeling techniques based on GraphSage, Relation GCN and Graph Attention Networks. Furthermore, the use of ensemble modeling increases model accuracy over a large range of prediction values. Trained on a large dataset of industrial circuits, the model achieves an average prediction R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> of 0.772 (110% better than XGBoost) and reduces average simulation errors from over 100% with designer's estimation to less than 10%.
7
Laurens van der Maaten, Geoffrey E. Hinton · Journal of Machine Learning Research · 2008 · 35.7K citations
Automatic differentiation in PyTorch
Adam Paszke, Sam Gross, Soumith Chintala et al. · 2017 · 11.1K citations
Aditya Grover, Jure Leskovec · 2016 · 10.9K citations
Deep Graph Library: Towards Efficient and Scalable Deep Learning on Graphs
Minjie Wang, Lingfan Yu, Da Zheng et al. · arXiv (Cornell University) · 2019 · 453 citations · Full text