Trax solver on Zynq with Deep Q-Network

Naru Sugimoto; Takuji Mitsuishi; Takahiro Kaneda; Chiharu Tsuruta; Ryotaro Sakai; Hideki Shimura; Hideharu Amano

doi:10.1109/FPT.2015.7393122

Trax solver on Zynq with Deep Q-Network

Naru Sugimoto, Takuji Mitsuishi, Takahiro Kaneda, Chiharu Tsuruta, Ryotaro Sakai, Hideki Shimura, Hideharu Amano

Department of Information and Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

7 Citations (Scopus)

Abstract

A software/hardware co-design system for a Trax solver is proposed. Implementation of Trax AI is challenging due to its complicated rules, so we adopted an embedded system called Zynq (Zynq-7000 AP SoC) and introduced a High Level Synthesis (HLS) design. We also added Deep Q-Network, a machine learning algorithm, to the system for use as an evaluation function. Our solver automatically optimizes its own evaluation function through games with humans or other AIs. The implemented solver works with a 150-MHz clock on the Xilinx XC7Z020-CLG484 of a Digilent ZedBoard. A part of the Deep Q-Network job can be executed on the FPGA of the Zynq board more than 26 times faster than with ARM Coretex-A9 650-MHz software.

Original language	English
Title of host publication	2015 International Conference on Field Programmable Technology, FPT 2015
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	272-275
Number of pages	4
ISBN (Print)	9781467390910
DOIs	https://doi.org/10.1109/FPT.2015.7393122
Publication status	Published - 2016 Jan 25
Event	International Conference on Field Programmable Technology, FPT 2015 - Queenstown, New Zealand Duration: 2015 Dec 7 → 2015 Dec 9

Other

Other	International Conference on Field Programmable Technology, FPT 2015
Country/Territory	New Zealand
City	Queenstown
Period	15/12/7 → 15/12/9

ASJC Scopus subject areas

Computer Science Applications
Hardware and Architecture
Software

Access to Document

10.1109/FPT.2015.7393122

Cite this

Sugimoto, N, Mitsuishi, T, Kaneda, T, Tsuruta, C, Sakai, R, Shimura, H & Amano, H 2016, Trax solver on Zynq with Deep Q-Network. in 2015 International Conference on Field Programmable Technology, FPT 2015., 7393122, Institute of Electrical and Electronics Engineers Inc., pp. 272-275, International Conference on Field Programmable Technology, FPT 2015, Queenstown, New Zealand, 15/12/7. https://doi.org/10.1109/FPT.2015.7393122

@inproceedings{3f7382a529614ae0af713aca8a1a3a68,

title = "Trax solver on Zynq with Deep Q-Network",

abstract = "A software/hardware co-design system for a Trax solver is proposed. Implementation of Trax AI is challenging due to its complicated rules, so we adopted an embedded system called Zynq (Zynq-7000 AP SoC) and introduced a High Level Synthesis (HLS) design. We also added Deep Q-Network, a machine learning algorithm, to the system for use as an evaluation function. Our solver automatically optimizes its own evaluation function through games with humans or other AIs. The implemented solver works with a 150-MHz clock on the Xilinx XC7Z020-CLG484 of a Digilent ZedBoard. A part of the Deep Q-Network job can be executed on the FPGA of the Zynq board more than 26 times faster than with ARM Coretex-A9 650-MHz software.",

author = "Naru Sugimoto and Takuji Mitsuishi and Takahiro Kaneda and Chiharu Tsuruta and Ryotaro Sakai and Hideki Shimura and Hideharu Amano",

year = "2016",

month = jan,

day = "25",

doi = "10.1109/FPT.2015.7393122",

language = "English",

isbn = "9781467390910",

pages = "272--275",

booktitle = "2015 International Conference on Field Programmable Technology, FPT 2015",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

note = "International Conference on Field Programmable Technology, FPT 2015 ; Conference date: 07-12-2015 Through 09-12-2015",

}

TY - GEN

T1 - Trax solver on Zynq with Deep Q-Network

AU - Sugimoto, Naru

AU - Mitsuishi, Takuji

AU - Kaneda, Takahiro

AU - Tsuruta, Chiharu

AU - Sakai, Ryotaro

AU - Shimura, Hideki

AU - Amano, Hideharu

PY - 2016/1/25

Y1 - 2016/1/25

N2 - A software/hardware co-design system for a Trax solver is proposed. Implementation of Trax AI is challenging due to its complicated rules, so we adopted an embedded system called Zynq (Zynq-7000 AP SoC) and introduced a High Level Synthesis (HLS) design. We also added Deep Q-Network, a machine learning algorithm, to the system for use as an evaluation function. Our solver automatically optimizes its own evaluation function through games with humans or other AIs. The implemented solver works with a 150-MHz clock on the Xilinx XC7Z020-CLG484 of a Digilent ZedBoard. A part of the Deep Q-Network job can be executed on the FPGA of the Zynq board more than 26 times faster than with ARM Coretex-A9 650-MHz software.

AB - A software/hardware co-design system for a Trax solver is proposed. Implementation of Trax AI is challenging due to its complicated rules, so we adopted an embedded system called Zynq (Zynq-7000 AP SoC) and introduced a High Level Synthesis (HLS) design. We also added Deep Q-Network, a machine learning algorithm, to the system for use as an evaluation function. Our solver automatically optimizes its own evaluation function through games with humans or other AIs. The implemented solver works with a 150-MHz clock on the Xilinx XC7Z020-CLG484 of a Digilent ZedBoard. A part of the Deep Q-Network job can be executed on the FPGA of the Zynq board more than 26 times faster than with ARM Coretex-A9 650-MHz software.

UR - http://www.scopus.com/inward/record.url?scp=84963548969&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84963548969&partnerID=8YFLogxK

U2 - 10.1109/FPT.2015.7393122

DO - 10.1109/FPT.2015.7393122

M3 - Conference contribution

AN - SCOPUS:84963548969

SN - 9781467390910

SP - 272

EP - 275

BT - 2015 International Conference on Field Programmable Technology, FPT 2015

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - International Conference on Field Programmable Technology, FPT 2015

Y2 - 7 December 2015 through 9 December 2015

ER -

Trax solver on Zynq with Deep Q-Network

Abstract

Other

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this