[18p-D104-1~7] What is "neuromorphic hardware " ?

[18p-D104-2] Brain-inspired technology based on functional devices in LSI metal layers towards artificial neural networks

〇Takao Marukame¹, Radu Berdan¹, Kumiko Nomura¹, Yoshifumi Nishi¹ (1.Toshiba corp.)

Keywords:brain-inspired hardware, neuromorphic hardware, Spike-timing dependent plasticity (STDP)

Deep learning is an artificial intelligence field which has recently achieved impressive success in practical commercial applications. Artificial neural networks (NN) and neuromorphic engineering have attracted much attention due to their possibility of becoming a fundamental technology platform for deep learning. In this paper, we try to categorize and introduce neuromorphic technologies and their issues. Especially, we focus on NN operations based on memristive devices, which can be implemented in LSI metal layers, and discuss the possibility of realizing post conventional computer systems (non-von Neumann architectures) capable of autonomous learning.