DC Field | Value | Language |
---|---|---|
dc.contributor.author | YI, WOOSEOK | - |
dc.contributor.author | PARK, JUN KI | - |
dc.contributor.author | KIM, JAE JOON | - |
dc.date.accessioned | 2018-05-10T08:30:20Z | - |
dc.date.available | 2018-05-10T08:30:20Z | - |
dc.date.created | 2018-02-21 | - |
dc.date.issued | 2017-10-19 | - |
dc.identifier.uri | https://oasis.postech.ac.kr/handle/2014.oak/41876 | - |
dc.description.abstract | We present a Classification Restricted Boltzmann Machine (ClassRBM) hardware for embedded machines with on-chip learning capability. The RBM is a kind of the generative model, and has been used as one of the most popular feature extractors and image preprocessors. The ClassRBM is a variant of the RBM that is adapted to classification tasks. We propose the multi-Neuron-Per-Class (multi-NPC) voting scheme for improving accuracy of ClassRBM. We also show that the Contrastive Divergence (CD), which is one of the most popular algorithms to train RBM, has limitations in multi-NPC ClassRBM learning and propose a modified CD algorithm to overcome the limitation. Experimental results on FPGA flatform for MNIST datasets confirm that classification accuracy of the proposed algorithm is∼ 2.12% higher than the conventional CD. | - |
dc.publisher | ACM/IEEE | - |
dc.relation.isPartOf | International Symposium on Rapid System Prototyping (RSP) | - |
dc.relation.isPartOf | Proceedings of RSP | - |
dc.title | GeCo: Classification Restricted Boltzmann Machine Hardware for On-chip Learning | - |
dc.type | Conference | - |
dc.type.rims | CONF | - |
dc.identifier.bibliographicCitation | International Symposium on Rapid System Prototyping (RSP) | - |
dc.citation.conferenceDate | 2017-10-19 | - |
dc.citation.conferencePlace | KO | - |
dc.citation.title | International Symposium on Rapid System Prototyping (RSP) | - |
dc.contributor.affiliatedAuthor | YI, WOOSEOK | - |
dc.contributor.affiliatedAuthor | PARK, JUN KI | - |
dc.contributor.affiliatedAuthor | KIM, JAE JOON | - |
dc.description.journalClass | 1 | - |
dc.description.journalClass | 1 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.
library@postech.ac.kr Tel: 054-279-2548
Copyrights © by 2017 Pohang University of Science ad Technology All right reserved.