I conduct research in the field of machine learning. I have several specific areas of interest.
One-class classfication
One area I am interested in is one-class classification and anomaly detection. I am interested in using kernel methods to derive boundaries around a single-class. I am also interested in how these methods can be made to update incrementally. Kernel methods are seen as computationally complex if the dataset is large, so this can be advantageous. Another interest of mine is distributed learning. In particular, is it possible to learn the boundary around a single class using data that is split across a number of nodes that can only communicate limited information, rather than the actual data that they have.
Time series analysis
I am also interested in time-series analysis. My approach has been to use Taken's embedding and a kernel method to derive a model of cyclo-stationary data. The approach is able to work on univariate and multivariate data, and has been shown to be competitive with other state-of-the-art approaches.
Application to novel biometrics
Finally, I am interested in the application of machine learning methods to novel biometrics. Smart phones are ubiquitous, are usually carried at all times, and contain many sensors. This makes them ideal for gathering data about a person that can be used to derive some novel biometrics. I have examined gait recognition and location-based recognition.