Summary

2023

Session Number:C1L-1

Session:

Number:C1L-12

Integrated Architecture of ITCAM and Reservoir Computing Using LSI Manufacturing Variation

Kunito Yuta,  Ajiki Go,  Akiyama Sayaka,  Kumaki Takeshi,  

pp.387-390

Publication Date:2023-09-21

Online ISSN:2188-5079

DOI:10.34385/proc.76.C1L-12

PDF download (986.1KB)

Summary:
In recent years, robots and artificial intelligence are used everywhere in our lives. Most of the communication robots that are currently being developed as human companions respond in a uniform way or are made to learn acquiredly. We are developing human partnership robot memory circuit for artificial intelligence with an innate personality. To give original individuality for AI-based robots, we focus on subtle variations during LSI manufacturing and built a new architecture IRC (ITCAM-based Reservoir Computing) that combines ITCAM (Ternary Content Addressable Memory with Individuality) and reservoir computing. The ITCAM can be reflected LSI manufacturing variation memory search results. Subtle variations during LSI manufacturing are physically difficult to duplicate due to impurities inside the chip. It is confirmed by simulation that the output results of simple waveforms such as sine waves differed by using Xilinx Spartan6 xc6slx150 variations as weights in machine learning.