Architecture of an FPGA accelerator for LDA-based inference

研究成果: Conference contribution

抄録

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.

本文言語English
ホスト出版物のタイトルProceedings - 18th IEEE/ACIS International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing, SNPD 2017
編集者Hiroaki Hirata, Nomiya Hiroki, Teruhisa Hochin
出版社Institute of Electrical and Electronics Engineers Inc.
ページ357-362
ページ数6
ISBN(電子版)9781509055043
DOI
出版ステータスPublished - 2017 8月 29
イベント18th IEEE/ACIS International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing, SNPD 2017 - Kanazawa, Japan
継続期間: 2017 6月 262017 6月 28

出版物シリーズ

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

Other

Other18th IEEE/ACIS International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing, SNPD 2017
国/地域Japan
CityKanazawa
Period17/6/2617/6/28

ASJC Scopus subject areas

  • ソフトウェア
  • 人工知能
  • コンピュータ ネットワークおよび通信
  • ハードウェアとアーキテクチャ

フィンガープリント

「Architecture of an FPGA accelerator for LDA-based inference」の研究トピックを掘り下げます。これらがまとまってユニークなフィンガープリントを構成します。

引用スタイル