In-memory computing for accelerating deep neural networks and neuro-vector-symbolic architectures

Abstract

In the first part of the tutorial, team’s recent efforts in employing a computational memory architecture for performing inference of deep neural networks will be presented. First, the phase-change memory technology used as computational memory will be described. Next, the application of computational memory to neural network inference will be explained, and experimental results will be presented based on a state-of-the-art fully-integrated 64-core computational phase-change memory chip. Finally, the open-source toolkit (https://analog- ai.mybluemix.net/) to simulate inference and training of neural networks with computational memory will be presented.