Bio
Elliot Stein is a PhD student at the University of Southampton, where is a member of the MINDS Centre for Doctoral Training, Vision Learning and Control group, and the Agents Interactions and Complexity Group. This diverse exposure has allowed him to interact with and learn from experts across a wide range of quantitative fields within the scope of machine learning. Having an undergraduate degree in mathematics from the University of Bath, Elliot's work places emphasis on understanding and innovating upon the algorithms and mathematics that underpin this rapidly advancing field.
Research interests
Driven by a desire for more versatile and human-like learning capabilities in machines, Elliot's research is focused on empowering machine learning algorithms to learn from a changing data stream. His work confronts the challenge of "Catastrophic Forgetting", a prevalent issue experienced by even the most advanced algorithms, where training on new data (for instance, adding a new class to a trained classifier model) triggers a rapid and significant degradation in performance on previously learned classes.
Moreover, Elliot's research interests extend to deep anomaly detection. By flipping the typical philosophy of machine learning from recognising specific patterns to detecting when something is meaningfully unusual, interesting problems and methods arise that have applications far beyond anomaly detection itself.
Working concurrently in these two exciting domains offers a compelling opportunity for 'cross-pollination', since identifying anomalous events in data is crucial for adapting to change and continual learning.