Skip to main content
Events
  1. Queen Mary University of London
  2. Events
  3. Past events
  4. 2022
  5. Items

IADS/DERI Seminar on "Graphs and invariants in Matrix data, with applications to computational linguistics"

When: Thursday, January 20, 2022, 11:00 AM - 12:00 PM
Where: remote

Speaker: Dr Sanjaye Ramgoolam, Reader in Theoretical Physics

Graphs and invariants in Matrix data, with applications to computational linguistics
Dr Sanjaye Ramgoolam, Reader in Theoretical Physics

Please join us for our first seminar of the year.  Sanjaye will describe low-dimensional vector representations of large matrix data, derived using constructions from random matrix theory and permutation symmetry. The vector spaces are constructed using directed graphs corresponding to invariants of the permutation symmetry. They are equipped with a geometry arising from random matrix theory. Gaussianity in the matrix data is quantified using a framework similar to the application of quantum field theory (QFT) in particle physics. This is made possible by mathematical analogies between QFT and random matrix theory, which can be viewed as zero dimensional QFT. In his talk, he will describe concrete applications of the vector representations in the context of data from computational linguistics. Predicting word relations involving synonyms, antonyms, hypernyms and hyponyms are some of the computational tasks in these applications. This talk is based on an interdisciplinary collaboration between theoretical physicists and computer scientists.

This seminar is open to QMUL internal and external staff and will be held online via Zoom; details to join will be sent upon registration. (No registration required from QMUL staff who receive DERI newsletters).

About the DERI seminar series
The Digital Environment Research Institute hosts weekly seminars, every Thursday at 11:00 am (UK time), in which QMUL academics and external guests discuss research underpinned by data science and AI. The remit of the seminar series is broad and covers all disciplines and focus areas in which data-driven projects are generating new knowledge and pushing the boundaries of discovery.

Series organizer: Martin Benning, School of Mathematical Sciences

Back to top