Modeling Creativity
Abstract
In his PhD dissertation (PDF), Tom De Smedt has looked at creativity and the thought-provoking question if and how machines can be creative. Why do ants seem so purposeful and creative when they are cooperating to harvest food; or where did spiders learn to make such intricate webs, while nature is inherently “blind” and without purpose? How does creativity work in humans? How exactly do we think new thoughts and construct creative solutions, which we unconsciously apply to everyday problems such as luring the housecat down from a tree, or (rarely and with much more conscious effort) to come up with a new theory of physics? To answer such questions we used computational models of creativity.
To this end, we have in turn developed two open source tools: Pattern and NodeBox for OpenGL. NOGL generates 2D animations based on Python programming code. It is useful for example for agent-based modeling (e.g., to study virtual ecosystems), data visualization (e.g., to study commonsense networks) and small games. Pattern is a Python toolkit for data mining, multilingual natural language processing, sentiment analysis, machine learning, graph analysis and visualization.
The research was funded by the Industrial Research Fund and conducted at CLiPS (Computational Linguistics & Psycholinguistics Research Center, University of Antwerp) under the supervision of prof. Walter Daelemans and Lucas Nijs (EMRG).
Deliverables
References
De Smedt T. (2013). Modeling Creativity: Case Studies in Python. University Press Antwerp. ISBN 978-90-5718-260-0.
De Smedt T., Daelemans, W. (2012). Pattern for Python. Journal of Machine Learning Research, 13: 2063–2067
De Smedt T., Menschaert L. (2012). VALENCE: Affective visualisation using EEG. Digital Creativity, 23(3-4): 272-277.
De Smedt T., Lechat L., Daelemans, W. (2012). Generative art inspired by nature, in NodeBox. In: Di Chio, Cecilia (ed.) Applications of Evolutionary Computation, Part II, LNCS 6625: 264–272. Berlin: Springer.