Program description

The creative computing research program concerns itself with the computational modelling of cognitive processes in the visual, linguistic, and ludological domains. Research in the program is predominantly empirical and automatic; through machine learning, computers induce models from data. By manipulating the representation of the tasks to be learned, the experimenter can use the experiments to induce new theories (by using knowledge-free representations) as well as validate existing theories (by adopting theory-driven representations). Although research in the group fully adopts the entire toolbox of machine learning methods, there is also a longstanding home-grown interest in the usage and development of memory-based (example-based, k-nearest neighbour) learning algorithms.

In the visual domain, highlight interests are the recognition of the "fingerprint" of master painters, texture classification, object and face recognition, facial dynamics analysis, and the detection of outliers through dimension reduction and dissimilarity metrics. In the linguistic domain, work is performed in automatic machine translation, word sense disambiguation, paraphrasing, text analytics, and text correction. In the ludological domain (games studies), research focuses on opponent modeling, developing artificially intelligent computer players, and measuring entertainment in games.