Artificial Neural Microcircuits for use in Neuromorphic System Design
Artificial Neural Microcircuits for use in Neuromorphic System Design
复制标题
用于神经形态系统设计的人工神经微电路
DOI：
10.1162/isal_a_00581
复制
发表时间：
2023
期刊：
影响因子：
--
通讯作者：
Walter A
中科院分区：
文献类型：
--
作者：
Walter A
人工神经网络（ann）是应用最广泛的生物形态计算形式之一。然而，与生物神经系统不同的是，目前的趋势是人工神经网络在结构上是同质的。此外，这种结构同质性需要应用复杂的训练和学习工具来生成特定于应用的人工神经网络，容易出现过拟合等陷阱。本文提出了另一种方法，灵感来自于神经微电路在生物学中所起的作用，即有机神经系统的所谓“基本处理元素”。如何使用人工神经微电路组装大型神经网络，作为现成的组件，是明确的；在展示通过新颖性检索制作此类微电路目录的初步工作的结果之前。
Artificial Neural Networks (ANNs) are one of the most widely employed forms of biomorphic computation. However (unlike the biological nervous systems they draw inspiration from) the current trend is for ANNs to be structurally homogeneous. Furthermore, this structural homogeneity requires the application of complex training & learning tools that produce application specific ANNs, susceptible to pitfalls like overfitting. In this paper, an alternative approach is suggested, inspired by the role played in biology by Neural Microcircuits, the so called “fundamental processing elements” of organic nervous systems. How large neural networks can be assembled using Artificial Neural Microcircuits, intended as off-the-shelf components, is articulated; before showing the results of initial work to produce a catalogue of such Microcircuits though the use of Novelty Search.