[C56] TraiNDSim: A Simulation Framework for Comprehensive Performance Evaluation of Neuromorphic Devices for On-Chip Training

Abstract

The advancement of neuromorphic devices (NDs) for processing deep neural networks has narrowed the accuracy gap with software-trained models. To accurately assess ND performance, reliable simulation frameworks for on-chip training are crucial. We critically evaluated existing frameworks, identifying key defects in the training process. Consequently, we introduce TraiNDSim, a novel framework that addresses these issues. In refining the training process, we propose an advanced conductance normalization strategy called layer-wise normalization, which limits the weight range by taking the initial weight distribution into account. Additionally, our framework integrates three conductance models, notably refining one of the conventional models to depend solely on nonlinearity. Moreover, it features a bi-directional weight representation method with a unique conductance compensation technique. Our comprehensive analysis using TraiNDSim demonstrates its effectiveness in accurately reflecting the impact of ND parameters on training, promising more precise device performance evaluations.