MSCA Postdoctoral Fellowship 2024: Expression of Interest

Description

Norwegian Centre of Excellence on Knowledge-driven Machine Learning Integreat announces an open call for selection of candidates for Marie Skłodowska-Curie Postdoctoral Fellowships 2024. 

We are searching for PhD holders (up to 8 years of from the date of award), who have not resided in Norway for more than 12 months during last 3 years at the time of the deadline (11 Sep 2024).

We welcome especially cross-disciplinary projects across Machine Learning (ML), Statistics, Logic, Language Technology and Ethics. 

You will join a vibrant community of scientists, from young talents to established researchers, across Machine Learning, Statistics, Logic, Language Technology and Ethics from Norway and internationally. Integreat’s aim is to train a community of early career researchers, to become the next generation of scientists of modern knowledge-driven machine learning. Applicants can connect to the existing projects at Integreat.

Required expertise

• PhD in an Integreat related research field
• Specialized expertise
• Visionary Leadership: Proven ability to lead and inspire researchers towards groundbreaking discoveries

Available projects (call 2024)

You may apply with your project, or with the project under.

Uncertainty quantification in the presence of logical constraints 

One of the most exciting frameworks to quantify uncertainty of predictions is conformal prediction (CP). Under appropriate exchangeability conditions, it provides a confidence set (credibility set if prediction is Bayesian) for a multivariate estimate with statistical coverage guarantees. This project will develop new CP methods for knowledge graphs (KGs), which are one of the most popular approaches for (semi-)structured data. There are many learning tasks that are based on KGs, such as KG completion, link prediction, and node and graph classification. Graph Neural Networks (GNNs) are very successful for learning on KGs and solving the mentioned tasks, but also have great potential for incorporating symbolic knowledge due to strong connections between GNNs and logics. We will develop methods for logic-aware CP on KGs using GNN for prediction and design new algorithms with theoretical guarantees. Then we will verify practical applicability and usefulness of these ideas and algorithms on benchmarks and on challenging real-world settings, with the possibility to develop industrial standard. This project will be at the interface between statistics, logic and machine learning.

Learning from human preferences in Large Language Models

A key step to optimise and control the behaviour of Large Language Models (LLMs) is to align them with curated human feedback. In this fine tuning phase, the LLM is retrained on the basis of a set of data of the type (prompt/input, answer1, answer2, preferred answer), where humans are asked to prefer one of the two possible answers to the given input. Humans can be substituted by algorithms. Reinforcement Learning with Human Feedback (RLHF) was introduced to improve the alignment of LLMs with human preferences (Ouyang et al., 2022). More recently DPO (Rafailov et al., 2023) has been  suggested, which avoids the reward model and Reinforcement Learning-based optimization. Learning from human preferences is a form of preference learning from incomplete ranking data. In this context, the mostly used method in LLM fine tuning is the Bradley Terry model. I this project we will explore alternative approaches, including Bayesian ones, with the purpose of preparing the data for fine tuning n a more precise way, as they represent the human (or algorithmical) preferences better.

Application procedure

Send your expression of interest to admin@integreat.no (running deadline), include:

• Letter of motivation and your research CV, including relevant details about your skills and experience, and your contact details.
• Short project description (max 1 page) and how your project may fit Integreat

All grant proposals will be written in close collaboration with a selected supervisor, and you will be provided with professional assistance from Integreat administration. Integreat supports of up to two project proposals per call. 

Contract duration

MSCA European Postdoctoral Fellowships are for up to two* years, while the recommended policy at UiO is three-year postdoctoral positions. Successfully granted MSCA fellows of the European Postdoctoral Fellowship with The Faculty of Mathematics and Natural Sciences as host institution, will be offered an additional year funded by the faculty, provided that they have applied for 24 months funding and can be employed in a postdoctoral fellowship position (*2,5 years for those including a non-academic placement period).

What can be funded?

The MSCA Postdoctoral Fellowship programme is open to all domains of research and innovation. The project can be designed freely by the applicant together with the host in a fully 'bottom-up' manner. For this call we are primarily interested in topics linked to Integreat research ambitions. 

What does the funding cover?

The grant provides an allowance to cover your living, travel and family costs. The grant is awarded to your host organisation (the return organisation for Global fellowships). The research costs and overheads of the host organisation(s) are also supported.

Anticipated commencement: 1 June 2025

Contact

Integreat Administrative Leader Maria Dikova.