Artificial Neural Networks and Machine Learning - ICANN 2013
Artificial Neural Networks and Machine Learning - ICANN 2013
复制标题
人工神经网络和机器学习 - ICANN 2013
DOI：
10.1007/978-3-642-40728-4_40
复制
发表时间：
2013
期刊：
影响因子：
--
通讯作者：
Alva P
中科院分区：
文献类型：
--
作者：
Alva P
在这项研究中，在基因型的两种形态表征，确定性和非确定性表征，比较了进化神经元形态的模式识别任务。确定性方法将树突形态明确地表示为一组可以产生单一表型的基因型分区。本研究采用的非确定性方法仅编码可产生多种表型的基因型中的分支概率。主要结果是，非确定性方法促使选择更对称的树突形态，这是在确定性方法中没有观察到的。
In this study, two morphological representations in the genotype, a deterministic and a nondeterministic representation, are compared when evolving a neuronal morphology for a pattern recognition task. The deterministic approach represents the dendritic morphology explicitly as a set of partitions in the genotype which can give rise to a single phenotype. The nondeterministic method used in this study encodes only the branching probability in the genotype which can produce multiple phenotypes. The main result is that the nondeterministic method instigates the selection of more symmetric dendritic morphologies which was not observed in the deterministic method.