iCASE PhD Studentship with QinetiQ via Newcastle University¶
This EPSRC iCASE collaboration investigates how auxiliary knowledge can support the assurance of large language models. The project studies ways to identify hallucinations, falsehoods, contradictions, and missing knowledge in LLM outputs while also improving model correctness through inference, training, and architecture research.
Project Details¶
| Item | Detail |
|---|---|
| Type | Studentship |
| Status | Live |
| Funder | Engineering and Physical Sciences Research Council |
| Value | £98,000 |
| Dates | 1 January 2024 - 31 December 2027 |
| Partner | QinetiQ / Newcastle University |
| Koorosh's role | Co-Investigator / supervisor |
Project Focus¶
- Knowledge-enhanced assurance for LLM outputs.
- Detection of hallucination, contradiction, falsehood, and missing knowledge.
- Improving LLM correctness through neuro-symbolic and knowledge-informed methods.
- Supporting trustworthy LLM use in high-assurance settings.