Emerging Nanoscale Devices for Neuromorphic Computing: from Devices, Modeling, Circuits and Emulation, to Cognitive Computing Applications
Abstract
Emerging nanoscale devices are promising candidates for high density in-memory computing paradigms, such as neuromorphic computing. The memristor device, introduced by HP labs 10 years ago, but predicted from theoretical principles by Chua 40 years ago, is slowly overcoming many technical difficulties to make it ready for massive use in neuromorphic systems. In this special session we focus on memristor-like devices for use in spiking event-driven systems. We cover from papers proposing alternative flavor devices, like the magnetoresistive device from CNRS or the ferroelectric based device from Bordeaux University, both specially being developed for neuromorphic spiking event-driven systems, to the genuine memristive devices being developed by CEA-LETI in collaboration with ST-Microelectronics and which are very close to being commercialized and large volumes. Papers will also cover circuit design aspects for developing neurons and interconnecting them into systems. The special session also covers the study of higher level cognitive systems, which can be studied using digital memristor emulators, as those developed by the Univ. of Baleares in collaboration with Leon Chua. The special session will also address practical applications of nanoscale memristive-type devices, by showing how to perform a high-level cognitive visual task, such as scene context recognition, by exploiting sensory signals from already available spiking event-driven cameras, like the Dynamic Vision Sensor (DVS), and demonstrating scene analysis with a spiking event-driven deep neural network. The Special Session brings together top leading scientists covering the fields of memristive devices (CNRS, CEA-LETI, Bordeaux), the design of circuits for such emerging devices (all participants), the emulation of memristors for simulating large scale systems (Univ. Baleares and Univ. of Berkeley), and the practical application of high-level Cognitive Visual Systems (CSIC and Univ. of Sevilla).
Emerging nanoscale devices are promising candidates for high density in-memory computing paradigms, such as neuromorphic computing. The memristor device, introduced by HP labs 10 years ago, but predicted from theoretical principles by Chua 40 years ago, is slowly overcoming many technical difficulties to make it ready for massive use in neuromorphic systems. In this special session we focus on memristor-like devices for use in spiking event-driven systems. We cover from papers proposing alternative flavor devices, like the magnetoresistive device from CNRS or the ferroelectric based device from Bordeaux University, both specially being developed for neuromorphic spiking event-driven systems, to the genuine memristive devices being developed by CEA-LETI in collaboration with ST-Microelectronics and which are very close to being commercialized and large volumes. Papers will also cover circuit design aspects for developing neurons and interconnecting them into systems. The special session also covers the study of higher level cognitive systems, which can be studied using digital memristor emulators, as those developed by the Univ. of Baleares in collaboration with Leon Chua. The special session will also address practical applications of nanoscale memristive-type devices, by showing how to perform a high-level cognitive visual task, such as scene context recognition, by exploiting sensory signals from already available spiking event-driven cameras, like the Dynamic Vision Sensor (DVS), and demonstrating scene analysis with a spiking event-driven deep neural network. The Special Session brings together top leading scientists covering the fields of memristive devices (CNRS, CEA-LETI, Bordeaux), the design of circuits for such emerging devices (all participants), the emulation of memristors for simulating large scale systems (Univ. Baleares and Univ. of Berkeley), and the practical application of high-level Cognitive Visual Systems (CSIC and Univ. of Sevilla).