2019/11/08 by Michael Rotman, Rotman, Michael, Lior Wolf +1
Engineering · Physics and Astronomy · #FOS: Computer and information sciences #FOS: Electrical engineering #Low-power high-performance VLSI design #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Model Reduction and Neural Networks #Signal Processing (eess.SP) #VLSI and FPGA Design Techniques #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.1911.03053
openalex publication_date 2019/11/08 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
The manual design of analog circuits is a tedious task of parameter tuning\nthat requires hours of work by human experts. In this work, we make a\nsignificant step towards a fully automatic design method that is based on deep\nlearning. The method selects the components and their configuration, as well as\ntheir numerical parameters. By contrast, the current literature methods are\nlimited to the parameter fitting part only. A two-stage network is used, which\nfirst generates a chain of circuit components and then predicts their\nparameters. A hypernetwork scheme is used in which a weight generating network,\nwhich is conditioned on the circuit's power spectrum, produces the parameters\nof a primal RNN network that places the components. A differential simulator is\nused for refining the numerical values of the components. We show that our\nmodel provides an efficient design solution, and is superior to alternative\nsolutions.\n