Read full announcement and apply!
Description
This is an interdisciplinary project on the boundary between formal logic, reasoning, and knowledge representation on the one hand, and statistical, data driven methods on the other.
As machine learning systems become increasingly “black-box,” the need for high performance models that are provably interpretable has never been greater. This interdisciplinary project merges the rigorous precision of formal logic with the predictive power of statistical learning.
Current data-driven methods often struggle to incorporate domain knowledge or handle structured information. Building on existing methods, such as Logic Regression, we aim to move beyond simple models that predict outcomes, but discover and exploit the underlying “logical rules” of a system. Moreover, where these methods are currently limited to Boolean “on/off” features, this project will explore how to extend them to Knowledge Graphs. We will leverage state-of-the-art methods in computational logic that respect pre-defined domain-specific constraints, and at the same time, sharpen predictive power while being transparent. We will at the same time use uncertainty aware probabilistic techniques to handle inferential and predictive uncertainties. The aim of this research is a new class of “Logic-First” learners that are smaller, faster, and more trustworthy.
Project duration: three years. A fourth year may be considered and it will involve 25 % of other career-promoting work. Other career-promoting work may consist of teaching, supervi-sion, and/or research assistance. This is dependent upon the qualification of the applicant and the current needs of the department.
Project requirements
This page describes requirements specific to Project 2. Applicants must also satisfy the general qualification requirements, PhD programme admission requirements, and application requirements described in the main call for PhD fellowships.
Required qualifications
- Master’s degree in computer science, mathematics, statistics, or another relevant field.
- Courses taken at bachelor and/or master level in some of the following: Formal Logic, Logic in Computer Science, Knowledge Representation, Statistical inference, Computational statistics, Bayesian statistics, Optimization, Machine Learning, or similar courses
- Documented programming skills, for example in some of Java, Python, C, C++, C#, R, Julia, etc.
Preferred qualifications
The following qualifications are not required but will be considered an advantage in the evaluation process:
- Documented competence through courses or master thesis work, in knowledge representation, reasoning, formal logic, or knowledge graphs.
- Documented competence through courses or master thesis work, in statistics, data science, optimization, or machine learning.
Contact for scientific inquiries
For questions regarding the scientific content, research objectives, or project-specific requirements