Dr Liam Hodgkinson
DECRA Research Fellow
School of Mathematics and Statistics
30 Scholarly works
2 Projects
HIGHLIGHTS
2026
Journal article
Computationally assisted proof of a novel O(3)×O(10)-invariant Einstein metric on S12
DOI: 10.1112/jlms.704772025
Conference Proceedings
Models of Heavy-Tailed Mechanistic Universality
2025
Conference Proceedings
Temperature Optimization for Bayesian Deep Learning
2025
Journal article
ComFe: An Interpretable Head for Vision Transformers
2025
Journal article
Preserving Angles Improves Feature Distillation
2024
Research grants (ARC, NHMRC, MRFF)
Universal Model Selection Criteria for Scientific Machine Learning
2024
Research grants (other domestic)
Universal Model Selection Criteria for Scientific Machine Learning
RECENT SCHOLARLY WORKS
2024
Conference Proceedings
How many classifiers do we need?
2023
Conference Proceedings
Test Accuracy vs. Generalization Gap: Model Selection in NLP without Accessing Training or Testing Data
DOI: 10.1145/3580305.35995182023
Conference Proceedings
A Heavy-Tailed Algebra for Probabilistic Programming
2023
Conference Proceedings
Monotonicity and Double Descent in Uncertainty Estimation with Gaussian Processes
2023
Conference Proceedings
When are ensembles really effective?