Brain-Inspired Computing Exploiting Carbon Nanotube Fets And Resistive Ram: Hyperdimensional Computing Case Study
2018 IEEE INTERNATIONAL SOLID-STATE CIRCUITS CONFERENCE - (ISSCC)（2018）
摘要
We demonstrate an end-to-end brain-inspired hyperdimensional (HD) computing nanosystem, effective for cognitive tasks such as language recognition, using heterogeneous integration of multiple emerging nanotechnologies. It uses monolithic 3D integration of carbon nanotube field-effect transistors (CNFETs, an emerging logic technology with significant energy-delay product (EDP) benefit vs. silicon CMOS [1]) and Resistive RAM (RRAM, an emerging memory that promises dense non-volatile and analog storage [2]). Due to their low fabrication temperature ( 20,000 sentences (6.4 million characters) per language pair. 2. One-shot learning (i.e., learning from few examples) using one text sample (∼100,000 characters) per language. 3. Resilient operation (98% accuracy) despite 78% hardware errors (circuit outputs stuck at 0 or 1). Our HD nanosystem consists of 1,952 CNFETs integrated with 224 RRAM cells.
更多查看译文
关键词
language recognition,heterogeneous integration,monolithic 3D integration,carbon nanotube field-effect transistors,analog storage,low fabrication temperature,fine-grained connections,dense vertical connections,chip stacking,packaging approaches,back-end-of-line inter-layer,energy-efficient circuits,HD computing,approximate accumulation circuits,gradual RRAM reset operation,HD nanosystem,hyperdimensional computing case study,end-to-end brain-inspired hyperdimensional computing nanosystem,cognitive tasks,one-shot learning,pairwise classification,RRAM single-bit storage,random projection circuits,area-efficient circuits,back-end-of-line inter-layer vias,storage layers,nonvolatile storage,nanotechnologies,resistive RAM,carbon nanotube FETs,RRAM cells,CNFETs,European languages,silicon CMOS,energy-delay product,logic technology,temperature 250.0 degC
AI 理解论文
溯源树
样例
生成溯源树，研究论文发展脉络
Chat Paper
正在生成论文摘要