Developed a theory that enables optimal composition of multiple value functions to drive reliable behaviour and efficient off-policy adaptation to dynamically changing rewards

Reconciled contradicting hypotheses about the computation of the tail of the striatum (threat prediction error and action prediction error), providing a unifying normative account

Formulated a safety-efficiency trade-off in human reinforcement learning, proposing an uncertainty-based solution.

Designed and performed a Virtual Reality experiment to test the model and performed hierarchical Bayesian modeling of human behavior.

Developed embodied generative models learned from demonstrations (in PyTorch) for safer robots (Franka Emika Panda arm), implementing fault detection and correction while explaining human sensorimotor experiment findings

Theorised homeostasis after injury as planning under partial observability, explaining paradoxes and providing a novel perspective on existing literature.

Analyzed chronic pain patient high-dimensional brain data using graph-theoretic measures to predict pain and fatigue levels from insular cortical networks.

Algorithmic modelling of addiction as drug-induced hijacking of a hierarchical reinforcement learning algorithm

Developed a toolbox for quantifying synchronization in multi-variate time-series data.

Developed a computer vision pipeline to track movements from handheld video, securing £3,000 in funding for healthy subject and patient trials

Prepared a suite of mobile-based cognitive tests for pain patients whose behaviour can be modelled using reinforcement learning algorithms

Developed a multimodal deep learning model enriched with targeted features for early Alzheimer’s detection from speech, achieving a 6.25% improvement in accuracy.