> Markdown version of [/jobs/ext/4677453-postdoctoral-researcher-human-centered-interpretable-machine-learning](https://www.wearedevelopers.com/jobs/ext/4677453-postdoctoral-researcher-human-centered-interpretable-machine-learning). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Postdoctoral Researcher Human-Centered Interpretable Machine Learning

- **Company:** Liacs
- **Location:** Leiden, Netherlands (Remote available)
- **Salary:** €3,706 - €5,760
- **Contract:** Temporary contract
- **Skills:** Artificial Intelligence, Data Analysis, Computer Programming, Data Mining, Information Theory, Python (Programming Language), Machine Learning, Media Technology, Information Technology
- **Published:** November 12, 2026
- **Apply:** https://www.academictransfer.com/en/jobs/365521/postdoctoral-researcher-human-centered-interpretable-machine-learning-10fte/apply/

## About the Role

* A PhD degree (or near completion) in Computer Science, Statistics, Artificial Intelligence, Data Science, or a related field;
 * Excellent knowledge of and experience with machine learning and statistics, in particular probabilistic modelling;
 * Knowledge of and experience with information-theoretic learning (e.g., the MDL principle), interpretable models (e.g., rule sets, trees, subgroups), or Bayesian model selection is a strong plus;
 * Highly motivated to both perform fundamental research in interpretable machine learning and apply the developed methods to real-world applications;
 * Creative, 'making things work' mentality, independent, and communicative team player;
 * A track record of publications in leading machine learning or data mining venues and excellent academic writing skills;
 * Excellent programming skills (preferably in Python);
 * Excellent proficiency in English (oral and written).

## Description

Machine learning is increasingly used to support decisions in high-stake domains such as health care, where experts need to understand, verify, and act on what a model has learned. Post-hoc explanations of black-box models offer only a partial view: they approximate a model's behaviour rather than reveal it. We therefore develop models that are interpretable by construction, such as probabilistic rule sets, subgroup sets, and histogram trees, whose structure, predictions, and uncertainty can be inspected directly.

A key challenge is that the same data typically supports many models that perform almost equally well: the Rashomon set. Data alone cannot determine which of these models is most meaningful, plausible, or useful to a domain expert. Meanwhile, experts hold background knowledge that current learning algorithms cannot exploit, and they may wish to steer, constrain, or reject what an algorithm proposes.

In this project, you will develop theory and algorithms for human-centered interpretable machine learning: methods that incorporate background knowledge and expert feedback into model learning and selection. The minimum description length (MDL) principle provides a natural foundation by viewing learning as compression. It balances goodness of fit against model complexity without ad-hoc regularization and offers a principled way to encode prior knowledge. Possible directions include knowledge-informed model selection, characterising and navigating the Rashomon set, and interactive learning with experts in the loop. Within this scope there is ample room to shape the research agenda based on your own expertise and ideas.

You will be embedded in the Explanatory Data Analysis (EDA) group at the Leiden Institute of Advanced Computer Science. The group develops algorithms and theory that help people discover what matters in complex data, understand cause and effect, and make trustworthy predictions and decisions, building on machine learning, statistics, and information theory. More information about the group can be found at https://eda.liacs.nl/

What you will do

 * Conduct independent, fundamental research on human-centered interpretable machine learning, in particular learning with background knowledge and expert feedback;
 * Develop and evaluate methods in collaboration with domain experts, in particular in health care;
 * Publish and present scientific articles in leading journals and conferences;
 * Co-supervise PhD and MSc students;
 * Contribute to teaching activities.

Where you will work
The faculty is home to more than 2,700 staff members and over 6,000 students enrolled in nine bachelor's and fifteen master's programs. We are located at the heart of the Bio Science Park, one of Europe's largest science parks, where academia and industry come together.