Principal Researcher, Alan Turing Institute
3 papers at NeurIPS 2025
We highlight the susceptibility of existing unlearning methods to relearning attacks and analyze the characteristics of robust methods by leveraging the weight-space perspective.
We introduce a novel program synthesis approach to output world models of complex, non-gridworld domains by representing world models as products of programmatic experts.
A mutual information estimator based on recent vector copula theory, which explicitly disentangles the marginal distributions and dependence structure in estimation.
