Layer Ensemble Averaging for Improving Memristor-Based Artificial Neural Network Performance
This addresses inefficiencies in in-memory computation for AI hardware, offering a method to mitigate defects in emerging memory devices, though it is incremental as it builds on existing mapping techniques.
The paper tackles the problem of hardware non-idealities in memristor-based artificial neural networks by proposing layer ensemble averaging to map pre-trained networks to defective hardware, achieving near-software performance with an average multi-task classification accuracy improvement from 61% to 72%.
Artificial neural networks have advanced due to scaling dimensions, but conventional computing faces inefficiency due to the von Neumann bottleneck. In-memory computation architectures, like memristors, offer promise but face challenges due to hardware non-idealities. This work proposes and experimentally demonstrates layer ensemble averaging, a technique to map pre-trained neural network solutions from software to defective hardware crossbars of emerging memory devices and reliably attain near-software performance on inference. The approach is investigated using a custom 20,000-device hardware prototyping platform on a continual learning problem where a network must learn new tasks without catastrophically forgetting previously learned information. Results demonstrate that by trading off the number of devices required for layer mapping, layer ensemble averaging can reliably boost defective memristive network performance up to the software baseline. For the investigated problem, the average multi-task classification accuracy improves from 61 % to 72 % (< 1 % of software baseline) using the proposed approach.