Research Associate/SRA in Deep Learning Theory

This position will contribute to the research programme "Advancing Modern Data-Driven Robust AI", which is funded by UKRI through a Turing AI World-Leading Fellowship led by co-investigators Professor Zoubin Ghahramani (Department of Engineering) and Dr Ferenc Huszár (Department of Computer Science and Technology).

The programme's goal is to understand and improve modern machine learning methods primarily by casting them in a probabilistic, information theoretic, causal inference framework. More specifically, the programme is focussed on four areas: (1) Robustness; (2) Integrating symbolic and statistical frameworks; (3) Scalable probabilistic inference methods and (4) A Theory of Generalisation and Transfer Learning. For this position, preference will be to select applicants whose expertise is on the Theory of Generalisation and Transfer Learning in Deep Learning.

This RA/SRA will be based at the Department of Computer Science and Technology (affectionately known as the Computer Lab) and will work primarily with Dr Ferenc Huszár (Computer Laboratory) in collaboration with Professor Zoubin Ghahramani (Engineering Department) as well as other members of the Machine Learning Groups at these two departments.

The core responsibilities include planning and conducting research in alignment with one or multiple components of the research program, ideally focussing on the theory of generalisation and transfer in deep learning. Additional responsibilities include developing research objectives and proposals; presentations and publications; assisting with the supervision of research students; liaising and networking with colleagues and students; planning and organising research resources and workshops.
 

Team and Environment

This project spans two departments. This position will be based in the Computer Lab and will be embedded in the ML@CL (Machine Learning at the Computer Lab) group which includes Professor Neil Lawrence, Dr Carl Henrik and Dr Ferenc Huszár as well as several other research fellows and students. The SRA will collaborate with the Machine Learning Group to work alongside Professors Richard Turner, Carl Edward Rasmussen, Zoubin Ghahramani, David Krueger and Miguel Hernández-Lobato as well as several research fellows and students. It is expected the SRA will collaborate with researchers across the two departments.

Our group values an open and inclusive culture. Members of the research group will be encouraged to engage in activities aimed at widening participation in Machine Learning Research, for example by contributing to summer schools, mentoring applicants and students from a variety of backgrounds.
 

Expected Qualifications

• PhD in computer science/mathematics/engineering or similar.

• Strong programming experience in Python and knowledge of machine learning libraries/frameworks such as NumPy, PyTorch, JAX or Tensorflow.

• Prior experience in carrying out and publishing machine learning research research (typically at NeurIPS, ICML, ICLR, ALT, COLT, JMLR or similar) related to one of the focus areas mentioned above preferably in the theory of generalisation and transfer learning.

• Strong background in branches of mathematics relevant to deep learning theory, such as probability, information theory, optimisation, linear algebra and geometry.

Beneficial Qualifications

• Research experience and publication track record in the theory of deep learning (examples include studying inductive biases of learning algorithms, developing generalisation bounds, putting forward theories of why deep learning works well, empirical investigations of deep learning phenomena).

• Prior experience with larger scale and rigorous deep learning experiments such as empirical studies of deep learning generalisation.

• Good software engineering practices and experience producing high-quality, reproducible research code including unit tests, documentation.

• Experience mentoring more junior colleagues.

The funds are available from June 2022. We expect to hold interviews late July.

Appointment will be made at a level (research associate or senior research associate) depending on the qualification and experience of the candidate.

For informal enquiries, please contact Dr Ferenc Huszár: fh277@cam.ac.uk.

You will need to upload a full curriculum vitae (CV) and a 1-page covering letter outlining your relevant past experience, and include the contact details for 2 or 3 referees. If you upload any additional documents which have not been requested, we will not be able to consider these as part of your application.

Please quote reference NR31976 on your application and in any correspondence about this vacancy.

The University actively supports equality, diversity and inclusion and encourages applications from all sections of society.

The University has a responsibility to ensure that all employees are eligible to live and work in the UK.