Alexandr N. Gubin, Vladislav L. Litvinov, Felix V. Filippov, Oleg N. Zlobin
{"title":"Research of the Adaptive Subsampling Effectiveness for Image Preprocessing in Neural Network Models","authors":"Alexandr N. Gubin, Vladislav L. Litvinov, Felix V. Filippov, Oleg N. Zlobin","doi":"10.1109/CTS53513.2021.9562770","DOIUrl":null,"url":null,"abstract":"The features of the functioning of the neural layers of the retina in the processing of data from photoreceptors for transmission to the optic nerve have been investigated. A procedure for the implementation of dynamically changing pooling is proposed in order to reduce the dimension of the data being processed. The procedure is also useful in problems of finding relationships using the attention mechanism. As a result of the computer simulation, the effectiveness of the proposed approach is shown.","PeriodicalId":371882,"journal":{"name":"2021 IV International Conference on Control in Technical Systems (CTS)","volume":"180 2 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2021-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2021 IV International Conference on Control in Technical Systems (CTS)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/CTS53513.2021.9562770","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
The features of the functioning of the neural layers of the retina in the processing of data from photoreceptors for transmission to the optic nerve have been investigated. A procedure for the implementation of dynamically changing pooling is proposed in order to reduce the dimension of the data being processed. The procedure is also useful in problems of finding relationships using the attention mechanism. As a result of the computer simulation, the effectiveness of the proposed approach is shown.
京公网安备 11010802042870号
京ICP备2023020795号-1
分享
求助内容：
应助结果提醒方式：
扫码关注我们
