English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Resistive random-access memory (ReRAM) is a promising technology fordesigning hardware accelerators for deep neural network (DNN) inferencing.However, stochastic noise in ReRAM crossbars can degrade the DNN inferencingaccuracy. We propose the design and optimization of a high-performance,area-and energy-efficient ReRAM-based hardware accelerator to achieve robustDNN inferencing in the presence of stochastic noise. We make two key technicalcontributions. First, we propose a stochastic-noise-aware training method,referred to as ReSNA, to improve the accuracy of DNN inferencing on ReRAMcrossbars with stochastic noise. Second, we propose an information-theoreticalgorithm, referred to as CF-MESMO, to identify the Pareto set of solutions totrade-off multiple objectives, including inferencing accuracy, area overhead,execution time, and energy consumption. The main challenge in this context isthat executing the ReSNA method to evaluate each candidate ReRAM design isprohibitive. To address this challenge, we utilize the continuous-fidelityevaluation of ReRAM designs associated with prohibitive high computation costby varying the number of training epochs to trade-off accuracy and cost.CF-MESMO iteratively selects the candidate ReRAM design and fidelity pair thatmaximizes the information gained per unit computation cost about the optimalPareto front. Our experiments on benchmark DNNs show that the proposedalgorithms efficiently uncover high-quality Pareto fronts. On average, ReSNAachieves 2.57% inferencing accuracy improvement for ResNet20 on the CIFAR-10dataset with respect to the baseline configuration. Moreover, CF-MESMOalgorithm achieves 90.91% reduction in computation cost compared to the popularmulti-objective optimization algorithm NSGA-II to reach the best solution fromNSGA-II.

Multi-Objective Optimization of ReRAM Crossbars for Robust DNN Inferencing under Stochastic Noise