Parameter Estimation of Combinational Circuits Using Machine Learning Methods

This article presents contemporary approaches to the application of machine learning methods for the analysis of combinational circuit parameters. It outlines the main challenges associated with the use of traditional circuit analysis methods, such as high computational resource and time requirements. the emphasis is placed on the analysis and identification of issues with existing parameter evaluation methods. Additionally, various machine learning algorithms are introduced, including graph neural networks and transformers, which have demonstrated promising results in related fields. Special attention is also given to the analysis of existing datasets and benchmarks. Algorithms for the generation of combinational circuits, aimed at training neural networks, are discussed. the results of the research have the potential to assist ASIC developers and manufacturers optimize the design process and ensure more efficient use of computing resources.