Autumn School on Computational Creativity

Topics and Abstracts (2011)

Simon Colton:

Building Creative Systems - Practical, Philosophical and Societal issues

In Computational Creativity research, we study how to build software systems that can take on some of the creative responsibility in arts and science projects. This naturally raises practical programming issues, and we cover some AI, graphics and vision techniques which can be used to build the generative aspects of creative software. These are illustrated with applications to discovery tasks in pure mathematics (in particular via discussion of the HR automated mathematical theory formation system), and with applications to automated painting (in particular via discussion of The Painting Fool project - www.thepaintingfool.com). However, there is a general impression in society that software cannot be creative, and this has real impact on the value people ascribe to the artefacts produced by software, in particular in the visual arts. Hence, this raises philosophical issues, and requires us to deal with the perception of software in society. These aspects are also covered in the course. Working with practical applications of Computational Creativity research in pure mathematics, video game design, the visual arts and graphic design, we have developed seven guiding principles for the building of creative systems, which are covered in the final presentation of the course.

  • Introduction to Computational Creativity: Historical aspects, methodological issues, AI methods, The assessment of generated artefacts and systems, Philosophical aspects
  • Building an Automated Mathematician: The HR Automated Theory Formation system, Combining reasoning systems, Discovery tasks in pure mathematics
  • Building an Automated Painter: Practical Aspects: The Painting Fool automated artist, Evolutionary art, Context free design grammars, Constraint solving, Layer management, Non-Photorealistic rendering, 3D modelling, Video generation, Physical printing issues
  • Building an Automated Painter: Cultural Aspects: Celebrating creative acts, The role of interpretation, Perception of creativity in computational systems, The art world, Handing over creative responsibility
  • Philosophical and Formalisation Issues: Dismantling the Turing Test, Computational Creativity Theory, Ludic Computing
  • Guiding Principals of Building Creative Systems: Ever decreasing circles, Paradigms lost, The whole is more than a sum of the parts, Climbing the meta-mountain, The creativity tripod, Beauty is in the mind of the beholde

Please see here for details of the research we do in the Computational Creativity Group at Imperial College.

Some overview articles:

Some webpages:


Graeme Ritchie:

  • The Formal Description of Computational Creativity: This talk introduces some of the views of computational creativity that have appeared in the research literature, including Boden's very influential taxonomy and Wiggins' characterisation of creativity as search. We consider what is different about creative processes, and how to make such ideas precise.
  • The profile of a creative program: If we are to make empirical claims about the extent to which programs are being creative, then we have to state quite precisely what we would accept as creative activity. This talk presents a set of criteria which could be used to characterise a "creative" program's behaviour. Each criterion focuses on some facet of the program's performance and offers a possible measure of success in one narrow respect.
  • Generating verbal humour: This lecture is a review of work which has been done in the past twenty years on programs which can create simple humour based on language tricks. Most of the programs are on a very small scale, both in terms of the types of joke created and the quantity of output produced. We discuss some of the problems faced in this subarea, and some of the methodological guidelines that should be followed.
  • Incongruity resolution and humour: Perhaps the most influential and widely cited "theory" of humour during the past 40 years has been "incongruity resolution" (IR). It is also the humour theory that is most centrally cognitive and most oriented towards processing. This talk reviews some of the variants of IR that have been proposed, and reflects on why there seems to be so little connection between IR and computational humour.
  • Riddle-building by rule: One of the most developed forms of computational joke generation has been the creation of "punning riddles", a simple form of question-answer joke using word-play, suitable for children around the ages of 8 to 11 years. This talk traces this line of research from 1992 to the present, outlines the mechanisms used within these programs, and demonstrates one of the most developed programs. If time permits, the audience will have a chance to generate their own novel riddles on papers, using simplified versions of the rules used in the programs.

The webpage of "The Joking Computer" is here.


Carlo Strapparava:

Witty, Affective, Persuasive (and Possibly Deceptive) Natural Language Processing

Dealing with creative language and in particular with affective, persuasive and even humorous language has been considered outside the scope of computational linguistics. Nonetheless it is possible to exploit current NLP techniques starting some explorations about it. We briefly review some computational experience about these typical creative genres. In particular we will focus on research issues relevant to computational humor and how affective meanings are expressed in natural language. We will introduce techniques for affective content detection and generation. As far as persuasive language is concerned, after the introduction of a specifically tagged corpus (exploiting for example public political speeches), some techniques for persuasive lexicon extraction and for predicting persuasiveness of discourses are provided together with prospective scenarios of application. We conclude the course showing some explorations in the automatic recognition of deceptive language.


Paulo Urbano:

Swarm Paintings as a Creative Human-Machine Process

  • The first presentations will introduce swarm paintings, a technique for the computational creation of artistic work inspired by collective natural phenomena. Swarm paintings will be contextualized as creative systems.
  • The following presentations will be focused on teaching Netlogo programming language.
  • Finally, the school participants will be then invited to engage in a lab project and make their own swarm paintings.

Tony Veale:

Creative Information Retrieval

Information retrieval (IR) and figurative language processing (FLP) could scarcely be more different in their treatment of language and meaning. IR views language as an open-ended set of mostly stable signs with which texts can be indexed and retrieved, focusing more on a text’s potential relevance than its potential meaning. In contrast, FLP views language as a system of unstable signs that can be used to talk about the world in creative new ways. There is another key difference: IR is practical, scalable and robust, and in daily use by millions of casual users. FLP is neither scalable nor robust, and not yet practical enough to migrate beyond the lab. This paper thus presents a mutually beneficial hybrid of IR and FLP, one that enriches IR with new operators to enable the non-literal retrieval of creative expressions, and which also transplants FLP into a robust, scalable framework in which practical applications of linguistic creativity can be implemented.


Geraint A. Wiggins:

Computational Modelling of Music Cognition and Musical Creativity

Music has a status in the creative arts which differs from other art forms. First, it is much more heavily dependent on memory than the vast majority of visual and tactile art; and, second, except in some very specific and special cases, it makes no reference to the world as experienced by the listener. In fact, the experience of listening to music is irresistibly time-based, since there can be no going back to previous sections of a work, without necessarily fracturing the work's structure - and this is true whether the work is great art or the most trivial of functional music. Because of the lack of reference, music can only be remembered in terms of itself. It follows that, to study music as a scientific phenomenon, one needs to study musical memory. Doing so leads us to consider how creativity can arise, and, unless one is content to resort to unscientific mysticism, one is inescapably drawn to the conclusion that creativity is a function of memory. In these five presentations, I will present work on the modelling of musical memory, and show how those models can be used also for musical production. I will place the work in a cognitive context: that is to say, the models presented will be credible candidates as models of human cognition at a particular level of abstraction; and I will present empirical evidence to suggest that they do effectively model some aspects of human behaviour. Finally, I will place the whole in the context of a theory of consciousness that accounts for why models of musical memory might account for the inspiration of composers.

07.10.2013 - 08:15 Hannu Toivonen
31.08.2011 - 18:44 Hannu Toivonen