OptiGAN: generative Adversarial Networks for goal optimized sequence generation

Trung Le; Viet Huynh; Michael Papasimeon; Dinh Phung

doi:10.1109/IJCNN48605.2020.9206842

OptiGAN: generative Adversarial Networks for goal optimized sequence generation

Trung Le
, Viet Huynh
, Michael Papasimeon
, Dinh Phung

Department of Data Science & AI

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Research › peer-review

6 Citations (Scopus)

Abstract

One of the challenging problems in sequence generation tasks is the optimized generation of sequences with specific desired goals. Current sequential generative models mainly generate sequences to closely mimic the training data, without direct optimization of desired goals or properties specific to the task. We introduce OptiGAN, a generative model that incorporates both Generative Adversarial Networks (GAN) and Reinforcement Learning (RL) to optimize desired goal scores using policy gradients. We apply our model to text and real-valued sequence generation, where our model is able to achieve higher desired scores out-performing GAN and RL baselines, while not sacrificing output sample diversity.

Original language	English
Title of host publication	2020 International Joint Conference on Neural Networks (IJCNN), 2020 Conference Proceedings2020 International Joint Conference on Neural Networks, IJCNN 2020 - Proceedings
Editors	Asim Roy
Place of Publication	Piscataway NJ USA
Publisher	IEEE, Institute of Electrical and Electronics Engineers
Pages	4387-4394
Number of pages	8
ISBN (Electronic)	9781728169262
ISBN (Print)	9781728169279
DOIs	https://doi.org/10.1109/IJCNN48605.2020.9206842
Publication status	Published - 2020
Event	IEEE International Joint Conference on Neural Networks 2020 - Virtual, Glasgow, United Kingdom Duration: 19 Jul 2020 → 24 Jul 2020 https://ieeexplore.ieee.org/xpl/conhome/9200848/proceeding (Proceedings) https://wcci2020.org/ijcnn-sessions/ (Website)

Publication series

Name	Proceedings of the International Joint Conference on Neural Networks
Publisher	IEEE
ISSN (Print)	2161-4393
ISSN (Electronic)	2161-4407

Conference

Conference	IEEE International Joint Conference on Neural Networks 2020
Abbreviated title	IJCNN 2020
Country/Territory	United Kingdom
City	Virtual, Glasgow
Period	19/07/20 → 24/07/20
Internet address	https://ieeexplore.ieee.org/xpl/conhome/9200848/proceeding (Proceedings) https://wcci2020.org/ijcnn-sessions/ (Website)

Keywords

Generative Adversarial Networks
Policy Gradients
Reinforcement Learning
Sequential Data

Access to Document

10.1109/IJCNN48605.2020.9206842

Cite this

Le, T., Huynh, V., Papasimeon, M., & Phung, D. (2020). OptiGAN: generative Adversarial Networks for goal optimized sequence generation. In A. Roy (Ed.), 2020 International Joint Conference on Neural Networks (IJCNN), 2020 Conference Proceedings2020 International Joint Conference on Neural Networks, IJCNN 2020 - Proceedings (pp. 4387-4394). (Proceedings of the International Joint Conference on Neural Networks). IEEE, Institute of Electrical and Electronics Engineers. https://doi.org/10.1109/IJCNN48605.2020.9206842

Le, Trung ; Huynh, Viet ; Papasimeon, Michael et al. / OptiGAN : generative Adversarial Networks for goal optimized sequence generation. 2020 International Joint Conference on Neural Networks (IJCNN), 2020 Conference Proceedings2020 International Joint Conference on Neural Networks, IJCNN 2020 - Proceedings. editor / Asim Roy. Piscataway NJ USA : IEEE, Institute of Electrical and Electronics Engineers, 2020. pp. 4387-4394 (Proceedings of the International Joint Conference on Neural Networks).

@inproceedings{9bdf739830254a5bb5602a31e0635110,

title = "OptiGAN: generative Adversarial Networks for goal optimized sequence generation",

abstract = "One of the challenging problems in sequence generation tasks is the optimized generation of sequences with specific desired goals. Current sequential generative models mainly generate sequences to closely mimic the training data, without direct optimization of desired goals or properties specific to the task. We introduce OptiGAN, a generative model that incorporates both Generative Adversarial Networks (GAN) and Reinforcement Learning (RL) to optimize desired goal scores using policy gradients. We apply our model to text and real-valued sequence generation, where our model is able to achieve higher desired scores out-performing GAN and RL baselines, while not sacrificing output sample diversity.",

keywords = "Generative Adversarial Networks, Policy Gradients, Reinforcement Learning, Sequential Data",

author = "Trung Le and Viet Huynh and Michael Papasimeon and Dinh Phung",

year = "2020",

doi = "10.1109/IJCNN48605.2020.9206842",

language = "English",

isbn = "9781728169279",

series = "Proceedings of the International Joint Conference on Neural Networks",

publisher = "IEEE, Institute of Electrical and Electronics Engineers",

pages = "4387--4394",

editor = "Asim Roy",

booktitle = "2020 International Joint Conference on Neural Networks (IJCNN), 2020 Conference Proceedings2020 International Joint Conference on Neural Networks, IJCNN 2020 - Proceedings",

address = "United States of America",

note = "IEEE International Joint Conference on Neural Networks 2020, IJCNN 2020 ; Conference date: 19-07-2020 Through 24-07-2020",

url = "https://ieeexplore.ieee.org/xpl/conhome/9200848/proceeding, https://wcci2020.org/ijcnn-sessions/",

}

Le, T, Huynh, V, Papasimeon, M & Phung, D 2020, OptiGAN: generative Adversarial Networks for goal optimized sequence generation. in A Roy (ed.), 2020 International Joint Conference on Neural Networks (IJCNN), 2020 Conference Proceedings2020 International Joint Conference on Neural Networks, IJCNN 2020 - Proceedings. Proceedings of the International Joint Conference on Neural Networks, IEEE, Institute of Electrical and Electronics Engineers, Piscataway NJ USA, pp. 4387-4394, IEEE International Joint Conference on Neural Networks 2020, Virtual, Glasgow, United Kingdom, 19/07/20. https://doi.org/10.1109/IJCNN48605.2020.9206842

OptiGAN: generative Adversarial Networks for goal optimized sequence generation. / Le, Trung; Huynh, Viet; Papasimeon, Michael et al.
2020 International Joint Conference on Neural Networks (IJCNN), 2020 Conference Proceedings2020 International Joint Conference on Neural Networks, IJCNN 2020 - Proceedings. ed. / Asim Roy. Piscataway NJ USA: IEEE, Institute of Electrical and Electronics Engineers, 2020. p. 4387-4394 (Proceedings of the International Joint Conference on Neural Networks).

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Research › peer-review

TY - GEN

T1 - OptiGAN

T2 - IEEE International Joint Conference on Neural Networks 2020

AU - Le, Trung

AU - Huynh, Viet

AU - Papasimeon, Michael

AU - Phung, Dinh

PY - 2020

Y1 - 2020

N2 - One of the challenging problems in sequence generation tasks is the optimized generation of sequences with specific desired goals. Current sequential generative models mainly generate sequences to closely mimic the training data, without direct optimization of desired goals or properties specific to the task. We introduce OptiGAN, a generative model that incorporates both Generative Adversarial Networks (GAN) and Reinforcement Learning (RL) to optimize desired goal scores using policy gradients. We apply our model to text and real-valued sequence generation, where our model is able to achieve higher desired scores out-performing GAN and RL baselines, while not sacrificing output sample diversity.

AB - One of the challenging problems in sequence generation tasks is the optimized generation of sequences with specific desired goals. Current sequential generative models mainly generate sequences to closely mimic the training data, without direct optimization of desired goals or properties specific to the task. We introduce OptiGAN, a generative model that incorporates both Generative Adversarial Networks (GAN) and Reinforcement Learning (RL) to optimize desired goal scores using policy gradients. We apply our model to text and real-valued sequence generation, where our model is able to achieve higher desired scores out-performing GAN and RL baselines, while not sacrificing output sample diversity.

KW - Generative Adversarial Networks

KW - Policy Gradients

KW - Reinforcement Learning

KW - Sequential Data

UR - https://www.scopus.com/pages/publications/85093829239

U2 - 10.1109/IJCNN48605.2020.9206842

DO - 10.1109/IJCNN48605.2020.9206842

M3 - Conference Paper

AN - SCOPUS:85093829239

SN - 9781728169279

T3 - Proceedings of the International Joint Conference on Neural Networks

SP - 4387

EP - 4394

BT - 2020 International Joint Conference on Neural Networks (IJCNN), 2020 Conference Proceedings2020 International Joint Conference on Neural Networks, IJCNN 2020 - Proceedings

A2 - Roy, Asim

PB - IEEE, Institute of Electrical and Electronics Engineers

CY - Piscataway NJ USA

Y2 - 19 July 2020 through 24 July 2020

ER -

Le T, Huynh V, Papasimeon M, Phung D. OptiGAN: generative Adversarial Networks for goal optimized sequence generation. In Roy A, editor, 2020 International Joint Conference on Neural Networks (IJCNN), 2020 Conference Proceedings2020 International Joint Conference on Neural Networks, IJCNN 2020 - Proceedings. Piscataway NJ USA: IEEE, Institute of Electrical and Electronics Engineers. 2020. p. 4387-4394. (Proceedings of the International Joint Conference on Neural Networks). doi: 10.1109/IJCNN48605.2020.9206842

OptiGAN: generative Adversarial Networks for goal optimized sequence generation

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Cite this