PhD Studentship: Formal Verification of AI Interfaces

Project title:  Formal Verification of AI Interfaces  

Supervisory Team: Ekaterina Komendantskaya, Alessandro Bruni (IT University of Copenhagen, Denmark), Reynald Affeldt (AIST, Japan)

Project description:

The field of computing is facing a conundrum caused by a clash in two opposing trends: on the one hand, the growth and proliferation of machine learning (ML) in software, and on the other hand, ever-growing concerns that, with ML models being a black-box technology, the safety, security and explainability of software that uses ML diminish.

To address these concerns, we need tools and languages that can serve as safe interfaces to ML components. Such safe ML interfaces will allow to specify the desired properties of ML models, train ML models to satisfy such properties, and verify that these desired properties do in fact hold in the final artifact. For example, one language that supports the safe ML interfaces approach is the Haskell DSL Vehicle [1]; and one iconic application for safe ML interfaces is in verifying autonomous car controllers [1].

At the moment, we use the Coq proof assistant and the recent MathComp-Analysis library to study the formalization of Differentiable Logics that allow to specify certain safety properties of ML models, and then compile them down into loss functions for training. We are looking for a PhD applicant with keen interest in mathematics, logic and/or Coq programming to join this team, to extend the initial study of [2] to a richer language such as [1]. The conditions of this PhD funding come with no restrictions on nationality but assume that a successful PhD candidate will have a competitive CV.

Please forward this advertisement to any interested individuals, and address any questions to: e.komendantskaya@soton.ac.uk

[1] Matthew L. Daggitt, Wen Kokke, Robert Atkey, Natalia Slusarz, Luca Arnaboldi, Ekaterina Komendantskaya
Vehicle: Bridging the Embedding Gap in the Verification of Neuro-Symbolic Programs. CoRR abs/2401.06379 , 2024.
https://arxiv.org/abs/2401.

[2] R. Affeldt, A. Bruni, E. Komendantskaya, N. Slusarz, and K. Stark.
Taming Differentiable Logics with Coq Formalisation. In Interactive Theorem Proving (ITP) 2024, 2024.
https://arxiv.org/abs/2403.

If you wish to discuss any details of the project informally, please contact Ekaterina Komendantskaya, Cyber-Physical Systems Group Research Group, Email: e.komendantskaya@soton.ac.uk .

Entry Requirements

A very good undergraduate degree (at least a UK 2:1 honours degree, or its international equivalent).

Closing date: 31 December 2024.  Applications will be considered in the order that they are received, the position will be considered filled when a suitable candidate has been identified.

Funding: We offer a range of funding opportunities for both UK and international students, including Bursaries and Scholarships.  For more information please visit PhD Scholarships | Doctoral College | University of Southampton  Funding will be awarded on a rolling basis, so apply early for the best opportunity to be considered.

How To Apply

Apply online: Search for a Postgraduate Programme of Study (soton.ac.uk). Select programme type (Research), 2024/25, Faculty of Physical Sciences and Engineering, next page select “PhD Computer Science (Full time)”. In Section 2 of the application form you should insert the name of the supervisor Ekaterina Komendantskaya.

Applications should include:

Research Proposal

Curriculum Vitae

Two reference letters

Degree Transcripts/Certificates to date

For further information please contact: feps-pgr-apply@soton.ac.uk

Advert information