Unsupervised Doodling and Painting with Improved SPIRAL

John F. J. Mellor; Eunbyung Park; Yaroslav Ganin; Igor Babuschkin; Tejas Kulkarni; Dan Rosenbaum; Andy Ballard; Theophane Weber; Oriol Vinyals; S. M. Ali Eslami

Unsupervised Doodling and Painting with Improved SPIRAL

John F. J. Mellor, Eunbyung Park, Yaroslav Ganin, Igor Babuschkin, Tejas Kulkarni, Dan Rosenbaum, Andy Ballard, Theophane Weber, Oriol Vinyals, S. M. Ali Eslami

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

Abstract

We investigate using reinforcement learning agents as generative models of images (extending arXiv:1804.01118). A generative agent controls a simulated painting environment, and is trained with rewards provided by a discriminator network simultaneously trained to assess the realism of the agent's samples, either unconditional or reconstructions. Compared to prior work, we make a number of improvements to the architectures of the agents and discriminators that lead to intriguing and at times surprising results. We find that when sufficiently constrained, generative agents can learn to produce images with a degree of visual abstraction, despite having only ever seen real photographs (no human brush strokes). And given enough time with the painting environment, they can produce images with considerable realism. These results show that, under the right circumstances, some aspects of human drawing can emerge from simulated embodiment, without the need for external supervision, imitation or social cues. Finally, we note the framework's potential for use in creative applications.

Original language	English
Title of host publication	Machine Learning for Creativity and Design Workshop, Neural Information Processing Systems (NeurIPS)
State	Published - 2 Oct 2019
Externally published	Yes

Bibliographical note

See https://learning-to-paint.github.io for an interactive version of this paper, with videos

Keywords

cs.CV
cs.LG
stat.ML
I.2; I.4

Access to Document

https://neurips2019creativity.github.io/doc/improved_spiral.pdf

Cite this

Mellor, J. F. J., Park, E., Ganin, Y., Babuschkin, I., Kulkarni, T., Rosenbaum, D., Ballard, A., Weber, T., Vinyals, O., & Eslami, S. M. A. (2019). Unsupervised Doodling and Painting with Improved SPIRAL. In Machine Learning for Creativity and Design Workshop, Neural Information Processing Systems (NeurIPS) https://neurips2019creativity.github.io/doc/improved_spiral.pdf

@inproceedings{5edd97bea0d24df09a4361fc09894ed8,

title = "Unsupervised Doodling and Painting with Improved SPIRAL",

abstract = " We investigate using reinforcement learning agents as generative models of images (extending arXiv:1804.01118). A generative agent controls a simulated painting environment, and is trained with rewards provided by a discriminator network simultaneously trained to assess the realism of the agent's samples, either unconditional or reconstructions. Compared to prior work, we make a number of improvements to the architectures of the agents and discriminators that lead to intriguing and at times surprising results. We find that when sufficiently constrained, generative agents can learn to produce images with a degree of visual abstraction, despite having only ever seen real photographs (no human brush strokes). And given enough time with the painting environment, they can produce images with considerable realism. These results show that, under the right circumstances, some aspects of human drawing can emerge from simulated embodiment, without the need for external supervision, imitation or social cues. Finally, we note the framework's potential for use in creative applications. ",

keywords = "cs.CV, cs.LG, stat.ML, I.2; I.4",

author = "Mellor, {John F. J.} and Eunbyung Park and Yaroslav Ganin and Igor Babuschkin and Tejas Kulkarni and Dan Rosenbaum and Andy Ballard and Theophane Weber and Oriol Vinyals and Eslami, {S. M. Ali}",

note = "See https://learning-to-paint.github.io for an interactive version of this paper, with videos",

year = "2019",

month = oct,

day = "2",

language = "אנגלית",

booktitle = "Machine Learning for Creativity and Design Workshop, Neural Information Processing Systems (NeurIPS)",

}

TY - GEN

T1 - Unsupervised Doodling and Painting with Improved SPIRAL

AU - Mellor, John F. J.

AU - Park, Eunbyung

AU - Ganin, Yaroslav

AU - Babuschkin, Igor

AU - Kulkarni, Tejas

AU - Rosenbaum, Dan

AU - Ballard, Andy

AU - Weber, Theophane

AU - Vinyals, Oriol

AU - Eslami, S. M. Ali

N1 - See https://learning-to-paint.github.io for an interactive version of this paper, with videos

PY - 2019/10/2

Y1 - 2019/10/2

N2 - We investigate using reinforcement learning agents as generative models of images (extending arXiv:1804.01118). A generative agent controls a simulated painting environment, and is trained with rewards provided by a discriminator network simultaneously trained to assess the realism of the agent's samples, either unconditional or reconstructions. Compared to prior work, we make a number of improvements to the architectures of the agents and discriminators that lead to intriguing and at times surprising results. We find that when sufficiently constrained, generative agents can learn to produce images with a degree of visual abstraction, despite having only ever seen real photographs (no human brush strokes). And given enough time with the painting environment, they can produce images with considerable realism. These results show that, under the right circumstances, some aspects of human drawing can emerge from simulated embodiment, without the need for external supervision, imitation or social cues. Finally, we note the framework's potential for use in creative applications.

AB - We investigate using reinforcement learning agents as generative models of images (extending arXiv:1804.01118). A generative agent controls a simulated painting environment, and is trained with rewards provided by a discriminator network simultaneously trained to assess the realism of the agent's samples, either unconditional or reconstructions. Compared to prior work, we make a number of improvements to the architectures of the agents and discriminators that lead to intriguing and at times surprising results. We find that when sufficiently constrained, generative agents can learn to produce images with a degree of visual abstraction, despite having only ever seen real photographs (no human brush strokes). And given enough time with the painting environment, they can produce images with considerable realism. These results show that, under the right circumstances, some aspects of human drawing can emerge from simulated embodiment, without the need for external supervision, imitation or social cues. Finally, we note the framework's potential for use in creative applications.

KW - cs.CV

KW - cs.LG

KW - stat.ML

KW - I.2; I.4

M3 - פרסום בספר כנס

BT - Machine Learning for Creativity and Design Workshop, Neural Information Processing Systems (NeurIPS)

ER -

Unsupervised Doodling and Painting with Improved SPIRAL

Abstract

Bibliographical note

Keywords

Access to Document

Fingerprint

Cite this