Centre for Reliable Machine Learning
The Centre’s members are interested in both theory and applications of machine learning. Our particular areas of strength include methods and performance guarantees that require minimal assumptions about the data. We conduct research in statistical learning theory, conformal prediction, prediction with expert advice, and game-theoretic foundations of statistics, probability, and finance. The Centre was established as Computer Learning Research Centre in 1998 and re-named Centre for Reliable Machine Learning in 2020.
About
The major goals and objectives of the Centre are:
- Development of machine learning theory, including design of efficient algorithms and analysis of their properties.
- Serving as a centre for interdisciplinary research.
- Provision of consultancy and advice to external organizations.
The Centre maintains a strong team of researchers; in addition to resident core academic staff, the Centre has visiting professors from USA, Europe and Russia. Extra expertise is brought in by our associate members, including reinforcement learning, natural language processing, and cyber-physical systems. Our external members (including Leonid Levin, Glenn Shafer, and Vladimir Vapnik) are key contributors to machine learning and computer science in general (https://cml.rhul.ac.uk/people.html). The Centre has also established and run the Kolmogorov Lecture and Medal (http://kolmogorov.cml.rhul.ac.uk/).
Research
The current theoretical activities of the Centre include:
- training neural networks in statistical learning theory,
- conformal prediction and testing with applications to change detection,
- prediction of packs and probabilistic regression in prediction with expert advice,
- adapting the game-theoretic foundations of probability to testing assumptions of machine learning and detecting the need for retraining.
The Centre also is active in financial, industrial, and biomedical applications that include:
- pharmaceutical industry, in particular selection of chemical compounds for drug development,
- information security, in particular identifying family of bots,
- molecular biology, in particular protein-protein interactions,
- medical diagnostics, in particular treatment of depression.
For other projects, see https://cml.rhul.ac.uk/projects.html.
