期刊文献+

Energy-efficient reconfigurable AI processor

Energy-efficient reconfigurable AI processor
下载PDF
导出
摘要 High computational energy-efficiency and rapid real-timeresponse are the major concerns for applications of artificial intelligencein low-power mobile and Internet of Things deviceswith limited storage capacity. Due to the outstanding superiorityof less memory requirement, low computation overheadand negligible accuracy degradation, deep neural networkswith binary/ternary weights (BTNNs) have been widely adoptedto replace traditional full-precision neural networks.
作者 Nanjian Wu
出处 《Journal of Semiconductors》 EI CAS CSCD 2019年第5期5-5,共1页 半导体学报(英文版)
  • 相关文献

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部