Ferroelectric-enhanced analog accelerators for robust, ultra-low power edge intelligence

A future driven by artificial intelligence (AI)-enhanced systems is expected to make our life easier, safer, or simply more entertaining. Yet, achieving these expectations in the realm of smart and interconnected devices (internet of things, 6G wireless communication, smart health monitoring, etc.) requires solving crucial challenges in the computation of AI models. Most importantly, computing these models on classical processors requires excessive computing power. In this project, we propose to investigate extremely energy-efficient neuromorphic computing, i.e., computing systems inspired by our brain, based on novel ferroelectric devices that are able to mimic biological computing mechanisms. We will create a library of these devices, fabricated and modelled, and ready to be integrated into typical computer-aided design tools used to design microelectronic systems. This also paves the way to building unique, multi-faceted neuromorphic expertise in Finland.