Our campus home, the ICICS Building.

About us A new frontier for Bayesian statistics has emerged with the expansion of data in the form of images, video, text and sounds on the world wide web. In this frontier there are many new and exciting applications, including the design of search engines for multimedia information retrieval, browsing tools for digital databases and automatic annotation of images with text. In the information age, human processing capabilities are no longer sufficient for understanding the vast quantities of multimedia available on the web and we must resort to statistical models to help us synthesize and peruse the data we have created for our consumption. The statistical models used to describe these databases are also massive. We are in the realm of models with thousands or even millions of parameters. This motivates the use of information theoretic regularisation tools and Markov Chain Monte Carlo (MCMC). Via Bayesian methods, we can also incorporate a priori knowledge into our models to further increase their precision. The Statistical Multimedia Learning Group is focused on harnessing the latest developments in statistics, machine learning, natural language understanding, computer vision and cognitive psychology in order to construct and evaluate methods that can make sense the vast array of multimedia available at our fingertips. Our research group, led by Nando de Freitas, is located in the Laboratory for Computational Intelligence in the Institute for Computing, Information and Cognitive Systems (ICICS) on the scenic University of British Columbia campus.

The Laboratory for Computational Intelligence. Conducting research in the lab.