{"title":"Hierarchical temporal memory implementation on FPGA using LFSR based spatial pooler address space generator","authors":"Madis Kerner, K. Tammemäe","doi":"10.1109/DDECS.2017.7934577","DOIUrl":null,"url":null,"abstract":"Hierarchical temporal memory (HTM) is the model of the neocortex functionality, developed by Numenta, Inc. The level of implementation does cover only the subset of actual neocortex layers functionality, but, however, is sufficient to be useful in different domain areas e.g. for a novelty or anomaly detection. Numenta provides their implementation of the HTM for commercial or research purposes as a software solution. The purpose of this work is to investigate the feasibility of implementing the HTM algorithm partly or entirely on FPGA, providing the suitable building block for the resource limited cyber physical systems. The uniqueness of the provided solution is based on resource efficient Linear Feedback Shift Registers (LFSR) as connection address generators, as well as using a simple serial interface for inter-column communication.","PeriodicalId":330743,"journal":{"name":"2017 IEEE 20th International Symposium on Design and Diagnostics of Electronic Circuits & Systems (DDECS)","volume":"96 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2017-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"4","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2017 IEEE 20th International Symposium on Design and Diagnostics of Electronic Circuits & Systems (DDECS)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/DDECS.2017.7934577","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 4
Abstract
Hierarchical temporal memory (HTM) is the model of the neocortex functionality, developed by Numenta, Inc. The level of implementation does cover only the subset of actual neocortex layers functionality, but, however, is sufficient to be useful in different domain areas e.g. for a novelty or anomaly detection. Numenta provides their implementation of the HTM for commercial or research purposes as a software solution. The purpose of this work is to investigate the feasibility of implementing the HTM algorithm partly or entirely on FPGA, providing the suitable building block for the resource limited cyber physical systems. The uniqueness of the provided solution is based on resource efficient Linear Feedback Shift Registers (LFSR) as connection address generators, as well as using a simple serial interface for inter-column communication.