Joshua Weston

• PhD Candidate, School of Mathematics and Physics, Queen's University Belfast

My research sits at the intersection of time-domain astrophysics, machine learning, and the societal implications of large-scale scientific automation. In particular, I am interested in how algorithmic systems can be designed, monitored and interpreted to translate high-volume survey data into reliable, high-fidelity transient discoveries, while retaining meaningful human oversight.

My current research focuses on:

Machine Learning for Transient Discovery and Host Association: I develop and deploy ML pipelines for real–bogus classification and extragalactic transient–host matching in major sky surveys. For the Asteroid Terrestrial-impact Last Alert System (ATLAS), I retrained convolutional neural networks that reduced false positives in the data and significantly lowered human validation demands. I have also built software tools for the Vera C. Rubin Observatory's Legacy Survey of Space and Time Deep Drilling Fields, integrating archival data and applying interpretable ML to improve host galaxy identification for faint, high-redshift transients. Ongoing work focuses on early alert filtering via brokers such as Lasair and constructing robust transient samples.

AI, Citizen Science, and Responsible Survey Automation: As part of the Leverhulme Interdisciplinary Network on Algorithmic Solutions (LINAS), I examine how AI systems reshape discovery, expertise and public engagement in astronomy. I am developing a citizen-science framework to label early Rubin alerts and systematically compare expert and public classifications, quantifying their impact on model performance, bias and interpretability. This work situates large-survey automation within broader questions of transparent and sustainable human–AI collaboration in scientific research.

• 2022–present PhD Student, Astrophysics Research Centre, Queen's University Belfast

MENAFN03032026000199003603ID1110814029