Andrew Parry

Email: a.parry.1@research.gla.ac.uk

Website: parry-parry.github.io

Office:

Room 502, Level 5

Sir Alwyn Williams Building

School of Computing Science

Glasgow, G12 8QQ

ORCID iDhttps://orcid.org/0000-0001-5446-8328

Research title: Modelling Uncertainties in Neural Networks for Improved Defence against Adversarial Attacks

Research Summary

I obtained my BSc in Computing Science from the University of Glasgow in 2022. After a research placement I joined the IR group at Glasgow working on search. I have made contributions to the PyTerrier framework maintained at Glasgow which can be accessed from my website.

Main Research Interests

  • Neural Ranking Models
  • Adversarial Attacks in Search
  • Efficient Neural Training

My Research

I am interested in how the overparameterized language models that underpin neural search often display behaviour which diverges from a user's notion of relevance. I am working to expose these weaknesses whilst improving efficiency to make neural search a more viable option in industry.

Publications

List by: Type | Date

Jump to: 2023
Number of items: 1.

2023

Parry, A., Fröbe, M., MacAvaney, S. , Potthast, M. and Hagen, M. (2023) Analyzing Adversarial Attacks on Sequence-to-Sequence Relevance Models. In: 46th European Conference on Information Retrieval (ECIR2024), Glasgow, Scotland, 24-28 March 2024, (Accepted for Publication)

This list was generated on Sat May 4 02:21:31 2024 BST.
Number of items: 1.

Conference Proceedings

Parry, A., Fröbe, M., MacAvaney, S. , Potthast, M. and Hagen, M. (2023) Analyzing Adversarial Attacks on Sequence-to-Sequence Relevance Models. In: 46th European Conference on Information Retrieval (ECIR2024), Glasgow, Scotland, 24-28 March 2024, (Accepted for Publication)

This list was generated on Sat May 4 02:21:31 2024 BST.