Architecture of an FPGA accelerator for LDA-based inference

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Latent Dirichlet allocation (LDA) based topic inference is a data classification method, that is used efficiently for extremely large data sets. However, the processing time is very large due to the serial computational behavior of the Markov Chain Monte Carlo method used for the topic inference. We propose a pipelined hardware architecture and memory allocation scheme to accelerate LDA using parallel processing. The proposed architecture is implemented on a reconfigurable hardware called FPGA (field programmable gate array), using OpenCL design environment. According to the experimental results, we achieved maximum speed-up of 2.38 times, while maintaining the same quality compared to the conventional CPU-based implementation.

Original languageEnglish
Title of host publicationProceedings - 18th IEEE/ACIS International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing, SNPD 2017
EditorsHiroaki Hirata, Nomiya Hiroki, Teruhisa Hochin
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages357-362
Number of pages6
ISBN (Electronic)9781509055043
DOIs
Publication statusPublished - 2017 Aug 29
Event18th IEEE/ACIS International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing, SNPD 2017 - Kanazawa, Japan
Duration: 2017 Jun 262017 Jun 28

Publication series

NameProceedings - 18th IEEE/ACIS International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing, SNPD 2017

Conference

Conference18th IEEE/ACIS International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing, SNPD 2017
Country/TerritoryJapan
CityKanazawa
Period17/6/2617/6/28

Keywords

  • Data classification
  • Gibbs sampling
  • Latent Dirichlet allocation
  • Machine learning
  • OpenCL for FPGA

Fingerprint

Dive into the research topics of 'Architecture of an FPGA accelerator for LDA-based inference'. Together they form a unique fingerprint.

Cite this