干旱区生态环境知识资源中心(Arid): Sparsity through evolutionary pruning prevents neuronal networks from overfitting

Arid

DOI	10.1016/j.neunet.2020.05.007
	Sparsity through evolutionary pruning prevents neuronal networks from overfitting
	Gerum, Richard C.; Erpenbeck, Andre; Krauss, Patrick; Schilling, Achim
通讯作者	Schilling, A
来源期刊	NEURAL NETWORKS
ISSN	0893-6080
EISSN	1879-2782
出版年	2020
卷号	128 页码:305-312
英文摘要	Modern Machine learning techniques take advantage of the exponentially rising calculation power in new generation processor units. Thus, the number of parameters which are trained to solve complex tasks was highly increased over the last decades. However, still the networks fail - in contrast to our brain - to develop general intelligence in the sense of being able to solve several complex tasks with only one network architecture. This could be the case because the brain is not a randomly initialized neural network, which has to be trained from scratch by simply investing a lot of calculation power, but has from birth some fixed hierarchical structure. To make progress in decoding the structural basis of biological neural networks we here chose a bottom-up approach, where we evolutionarily trained small neural networks in performing a maze task. This simple maze task requires dynamic decision making with delayed rewards. We were able to show that during the evolutionary optimization random severance of connections leads to better generalization performance of the networks compared to fully connected networks. We conclude that sparsity is a central property of neural networks and should be considered for modern Machine learning approaches. (C) 2020 The Author(s). Published by Elsevier Ltd.
英文关键词	Evolution Artificial neural networks Maze task Evolutionary algorithm Overfitting Biological plausibility
类型	Article
语种	英语
开放获取类型	Other Gold
收录类别	SCI-E
WOS记录号	WOS:000567770800007
WOS关键词	PATH-INTEGRATION ; DESERT ANTS ; CATAGLYPHIS ; CONNECTOME ; MICROGLIA ; FEATURES ; SYSTEM ; EDGE ; MAP
WOS类目	Computer Science, Artificial Intelligence ; Neurosciences
WOS研究方向	Computer Science ; Neurosciences & Neurology
资源类型	期刊论文
条目标识符	http://119.78.100.177/qdio/handle/2XILL650/326168
作者单位	[Gerum, Richard C.] Friedrich Alexander Univ Erlangen Nurnberg FAU, Dept Phys, Biophys Grp, Nurnberg, Germany; [Erpenbeck, Andre] Tel Aviv Univ TAU, Sch Chem, Raymond & Beverley Sackler Ctr Computat Mol & Mat, Tel Aviv, Israel; [Krauss, Patrick; Schilling, Achim] Univ Hosp Erlangen, Neurosci Lab, Expt Otolaryngol, Erlangen, Germany; [Krauss, Patrick; Schilling, Achim] Friedrich Alexander Univ Erlangen Nurnberg FAU, Chair English Philol & Linguist, Cognit Computat Neurosci Grp, Erlangen, Germany; [Krauss, Patrick] Univ Groningen, Univ Med Ctr Groningen UMCG, Dept Otorhinolaryngol Head & Neck Surg, Groningen, Netherlands
推荐引用方式 GB/T 7714	Gerum, Richard C.,Erpenbeck, Andre,Krauss, Patrick,et al. Sparsity through evolutionary pruning prevents neuronal networks from overfitting[J],2020,128:305-312.
APA	Gerum, Richard C.,Erpenbeck, Andre,Krauss, Patrick,&Schilling, Achim.(2020).Sparsity through evolutionary pruning prevents neuronal networks from overfitting.NEURAL NETWORKS,128,305-312.
MLA	Gerum, Richard C.,et al."Sparsity through evolutionary pruning prevents neuronal networks from overfitting".NEURAL NETWORKS 128(2020):305-312.