An adjoint sensitivity technique for dynamic neural-network modeling and design of high-speed interconnect

摘要

In this article, we develop an adjoint dynamic neural network (ADNN) technique aimed at enhancing computer-aided design (CAD) of high-speed VLSI modules. A novel formulation for exact sensitivities is achieved by defining an adjoint of a dynamic neural network (DNN). We further present an in-depth description of how our ADNN is computationally linked with the original DNN in the transient-simulation environment in order to improve the efficiency of solving the ADNN. Using ADNN-enabled sensitivities, we develop a new training algorithm that facilitates DNN learning of nonlinear transients directly from continuous time-domain waveform data. The proposed algorithm is also expanded to enable physics-based nonlinear circuit CAD through faster sensitivity computations. Applications of our ADNN approach in transient modeling and circuit design are demonstrated by the examples of modeling physics-based high-speed interconnect drivers and gradient-based signal integrity optimization. (C) 2006 Wiley Periodicals, Inc.