首页> 外文期刊>電子情報通信学会技術研究報告. ニュ-ロコンピュ-ティング. Neurocomputing >Analog neural-network associative memories with symmetrically connected CMOS synapses
【24h】

Analog neural-network associative memories with symmetrically connected CMOS synapses

机译:模拟神经网络关联回忆,具有对称连接的CMOS突触

获取原文
获取原文并翻译 | 示例
           

摘要

Analog neural network hardware has many advantages such as quicker response and smaller chip area over digital or pulse hardware However, it is not widely investigated because it has been considered to have drawbacks in synaptic weight precision and noise In order to overcome the problems, we propose novel designs. (l) a quick-response synaptic weight with high symmetry in both small-signal and large-amplitude operation and (2) an open-loop, quick-response and parallel-symmetric neuron circuit that has inverse and non-inverse output with high symmetry in delay and amplitude. We report the fabrication of an associative memory chip and its characteristics.
机译:模拟神经网络硬件具有许多优点,如诸如数字或脉冲硬件的更快的响应和更小的芯片区域,但没有被广泛调查,因为它已被认为具有突触重量精度和噪声的缺点,以克服问题,我们提出 新颖的设计。 (l)在小信号和大幅度操作中具有高对称性的快速响应突触重量,(2)开环,快速响应和并联对称神经元电路,具有高的逆输出 延迟和幅度的对称性。 我们报告了关联存储器芯片的制造及其特性。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号