Machine Learning & Statistics

Insight researchers have extensive experience in mainstream research on machine learning and statistics that includes, supervised and unsupervised learning, dimension reduction, Bayesian methods and the application of these methods in a variety of domains. Our current work focuses on the next generation of machine learning and statistics algorithms that can operate on large-scale, dynamic data.

This may require real-time classification, datasets may be distributed or streamed, and data may exhibit highly dynamic characteristics, requiring adaptive algorithms. Dimensionality reduction is a vital aspect in such setting. It is not uncommon to have many thousands of predictor variables or feature vectors. The key challenge is to identify compact subsets which influence outcomes of interest. The objective of this strand is to develop novel algorithms and methods to derive insights from large scale real-world data sets.