Enabling Neuromorphic Computing for Multi-Tenant AI

In this Focus Group, Hans Fischer Fellow Prof. Chenchen Liu (University Maryland, Baltimore County) collaborates with Prof. Ulf Schlichtmann (Electronic Design Automation, TUM).

A distinctive trend in recent artificial intelligence (AI) applications is that they are evolving from singular tasks based on a single deep learning model – e.g., a deep neural network (DNN) – to complex “multi-tenant” scenarios with multiple DNN models being executed concurrently. The deployment of multi-tenant DNNs still presents significant difficulties. The goal of this research is to develop an innovative neuromorphic computing engine that can efficiently support multi-tenant AI. The neuromorphic engine not only can support complex multi-tenant DNNs computing with flexible resource and function configurations, but also can host model interactions across individual tenants’ computing instances with redefined multi-tenant data flow logistics and immediate computations. This research will benefit the computer system community at large by inspiring an interactive design philosophy between emerging complex AI applications, deep learning algorithms and their computing principles, and novel computing paradigms.