Learning Optimal Q-Function Using Deep Boltzmann Machine for Reliable Trading of Cryptocurrency

Seok Jun Bu, Sung-Bae Cho

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

The explosive price volatility from the end of 2017 to January 2018 shows that bitcoin is a high risk asset. The deep reinforcement algorithm is straightforward idea for directly outputs the market management actions to achieve higher profit instead of higher price-prediction accuracy. However, existing deep reinforcement learning algorithms including Q-learning are also limited to problems caused by enormous searching space. We propose a combination of double Q-network and unsupervised pre-training using Deep Boltzmann Machine (DBM) to generate and enhance the optimal Q-function in cryptocurrency trading. We obtained the profit of 2,686% in simulation, whereas the best conventional model had that of 2,087% for the same period of test. In addition, our model records 24% of profit while market price significantly drops by −64%.

Original languageEnglish
Title of host publicationIntelligent Data Engineering and Automated Learning – IDEAL 2018 - 19th International Conference, Proceedings
EditorsHujun Yin, Paulo Novais, David Camacho, Antonio J. Tallón-Ballesteros
PublisherSpringer Verlag
Pages468-480
Number of pages13
ISBN (Print)9783030034924
DOIs
Publication statusPublished - 2018 Jan 1
Event19th International Conference on Intelligent Data Engineering and Automated Learning, IDEAL 2018 - Madrid, Spain
Duration: 2018 Nov 212018 Nov 23

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11314 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other19th International Conference on Intelligent Data Engineering and Automated Learning, IDEAL 2018
CountrySpain
CityMadrid
Period18/11/2118/11/23

Fingerprint

Boltzmann Machine
Q-function
Profit
Profitability
Q-learning
Reinforcement learning
Reinforcement
Reinforcement Learning
Volatility
Learning algorithms
Learning Algorithm
Prediction
Output
Model
Learning
Electronic money
Simulation
Market

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Bu, S. J., & Cho, S-B. (2018). Learning Optimal Q-Function Using Deep Boltzmann Machine for Reliable Trading of Cryptocurrency. In H. Yin, P. Novais, D. Camacho, & A. J. Tallón-Ballesteros (Eds.), Intelligent Data Engineering and Automated Learning – IDEAL 2018 - 19th International Conference, Proceedings (pp. 468-480). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11314 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-03493-1_49
Bu, Seok Jun ; Cho, Sung-Bae. / Learning Optimal Q-Function Using Deep Boltzmann Machine for Reliable Trading of Cryptocurrency. Intelligent Data Engineering and Automated Learning – IDEAL 2018 - 19th International Conference, Proceedings. editor / Hujun Yin ; Paulo Novais ; David Camacho ; Antonio J. Tallón-Ballesteros. Springer Verlag, 2018. pp. 468-480 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{8ea4c420a39d48d0be0d09ee421370f2,
title = "Learning Optimal Q-Function Using Deep Boltzmann Machine for Reliable Trading of Cryptocurrency",
abstract = "The explosive price volatility from the end of 2017 to January 2018 shows that bitcoin is a high risk asset. The deep reinforcement algorithm is straightforward idea for directly outputs the market management actions to achieve higher profit instead of higher price-prediction accuracy. However, existing deep reinforcement learning algorithms including Q-learning are also limited to problems caused by enormous searching space. We propose a combination of double Q-network and unsupervised pre-training using Deep Boltzmann Machine (DBM) to generate and enhance the optimal Q-function in cryptocurrency trading. We obtained the profit of 2,686{\%} in simulation, whereas the best conventional model had that of 2,087{\%} for the same period of test. In addition, our model records 24{\%} of profit while market price significantly drops by −64{\%}.",
author = "Bu, {Seok Jun} and Sung-Bae Cho",
year = "2018",
month = "1",
day = "1",
doi = "10.1007/978-3-030-03493-1_49",
language = "English",
isbn = "9783030034924",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "468--480",
editor = "Hujun Yin and Paulo Novais and David Camacho and Tall{\'o}n-Ballesteros, {Antonio J.}",
booktitle = "Intelligent Data Engineering and Automated Learning – IDEAL 2018 - 19th International Conference, Proceedings",
address = "Germany",

}

Bu, SJ & Cho, S-B 2018, Learning Optimal Q-Function Using Deep Boltzmann Machine for Reliable Trading of Cryptocurrency. in H Yin, P Novais, D Camacho & AJ Tallón-Ballesteros (eds), Intelligent Data Engineering and Automated Learning – IDEAL 2018 - 19th International Conference, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11314 LNCS, Springer Verlag, pp. 468-480, 19th International Conference on Intelligent Data Engineering and Automated Learning, IDEAL 2018, Madrid, Spain, 18/11/21. https://doi.org/10.1007/978-3-030-03493-1_49

Learning Optimal Q-Function Using Deep Boltzmann Machine for Reliable Trading of Cryptocurrency. / Bu, Seok Jun; Cho, Sung-Bae.

Intelligent Data Engineering and Automated Learning – IDEAL 2018 - 19th International Conference, Proceedings. ed. / Hujun Yin; Paulo Novais; David Camacho; Antonio J. Tallón-Ballesteros. Springer Verlag, 2018. p. 468-480 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11314 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Learning Optimal Q-Function Using Deep Boltzmann Machine for Reliable Trading of Cryptocurrency

AU - Bu, Seok Jun

AU - Cho, Sung-Bae

PY - 2018/1/1

Y1 - 2018/1/1

N2 - The explosive price volatility from the end of 2017 to January 2018 shows that bitcoin is a high risk asset. The deep reinforcement algorithm is straightforward idea for directly outputs the market management actions to achieve higher profit instead of higher price-prediction accuracy. However, existing deep reinforcement learning algorithms including Q-learning are also limited to problems caused by enormous searching space. We propose a combination of double Q-network and unsupervised pre-training using Deep Boltzmann Machine (DBM) to generate and enhance the optimal Q-function in cryptocurrency trading. We obtained the profit of 2,686% in simulation, whereas the best conventional model had that of 2,087% for the same period of test. In addition, our model records 24% of profit while market price significantly drops by −64%.

AB - The explosive price volatility from the end of 2017 to January 2018 shows that bitcoin is a high risk asset. The deep reinforcement algorithm is straightforward idea for directly outputs the market management actions to achieve higher profit instead of higher price-prediction accuracy. However, existing deep reinforcement learning algorithms including Q-learning are also limited to problems caused by enormous searching space. We propose a combination of double Q-network and unsupervised pre-training using Deep Boltzmann Machine (DBM) to generate and enhance the optimal Q-function in cryptocurrency trading. We obtained the profit of 2,686% in simulation, whereas the best conventional model had that of 2,087% for the same period of test. In addition, our model records 24% of profit while market price significantly drops by −64%.

UR - http://www.scopus.com/inward/record.url?scp=85057076158&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85057076158&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-03493-1_49

DO - 10.1007/978-3-030-03493-1_49

M3 - Conference contribution

AN - SCOPUS:85057076158

SN - 9783030034924

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 468

EP - 480

BT - Intelligent Data Engineering and Automated Learning – IDEAL 2018 - 19th International Conference, Proceedings

A2 - Yin, Hujun

A2 - Novais, Paulo

A2 - Camacho, David

A2 - Tallón-Ballesteros, Antonio J.

PB - Springer Verlag

ER -

Bu SJ, Cho S-B. Learning Optimal Q-Function Using Deep Boltzmann Machine for Reliable Trading of Cryptocurrency. In Yin H, Novais P, Camacho D, Tallón-Ballesteros AJ, editors, Intelligent Data Engineering and Automated Learning – IDEAL 2018 - 19th International Conference, Proceedings. Springer Verlag. 2018. p. 468-480. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-030-03493-1_49