GAIG Game AI Research Group @ QMUL

Bridging Generative Deep Learning and Computational Creativity


Abstract

We aim to help bridge the research fields of generative deep learning and computational creativity by way of the creative AI community, and to advocate the common objective of more creatively autonomous generative learning systems. We argue here that generative deep learning models are inherently limited in their creative abilities because of a focus on learning for perfection. To highlight this, we present a series of techniques which actively diverge from standard usage of deep learning, with the specific intention of producing novel and interesting artefacts. We sketch out some avenues for improvement of the training and application of generative models and discuss how previous work on the evaluation of novelty in a computational creativity setting could be harnessed for such improvements. We end by describing how a two-way bridge between the research fields could be built.
URL: http://computationalcreativity.net/iccc20/papers/164-iccc20.pdf

Cite this work

@inproceedings{berns2020bridging,
author= {Berns, Sebastian and Colton, Simon},
title= {{Bridging Generative Deep Learning and Computational Creativity}},
year= {2020},
booktitle= {{Proceedings of the 11th International Conference on Computational Creativity (ICCC’20)}},
url= {http://computationalcreativity.net/iccc20/papers/164-iccc20.pdf},
abstract= {We aim to help bridge the research fields of generative deep learning and computational creativity by way of the creative AI community, and to advocate the common objective of more creatively autonomous generative learning systems. We argue here that generative deep learning models are inherently limited in their creative abilities because of a focus on learning for perfection. To highlight this, we present a series of techniques which actively diverge from standard usage of deep learning, with the specific intention of producing novel and interesting artefacts. We sketch out some avenues for improvement of the training and application of generative models and discuss how previous work on the evaluation of novelty in a computational creativity setting could be harnessed for such improvements. We end by describing how a two-way bridge between the research fields could be built.},
}

Similar Posts

Comments