Publications & ongoing work
Research
Work on machine learning optimisation, model alignment, and the reliability of systems that must know when not to answer.
01
PublishedFirst AuthorarXiv2025
Enhancing Speech Emotion Recognition via Fine-Tuning Pre-Trained Models and Hyper-Parameter Optimisation
Research into resource-efficient AutoML and hyperparameter optimisation for speech emotion recognition, using pretrained representations, Bayesian optimisation, and TPE.
- Speech emotion recognition
- Fine-tuning
- AutoML
- Bayesian optimisation
02
Manuscript in preparationReinforcement Learning Lead
Model Alignment for Reliable LLM Abstention
Research on model uncertainty and reliable refusal or abstention behaviour, with work spanning reinforcement learning, GRPO, DPO, and fine-tuning.
- LLM reliability
- Abstention
- GRPO
- DPO
- Fine-tuning
Paper, code, and project links will be added when verified URLs are supplied.