Matthijs van der Lende

I am currently a Master’s student in Artificial Intelligence at the University of Groningen. My research interests is broadly about how machine-learning systems learn, adapt, and make decisions, with a focus on continual learning, reinforcement learning, probabilistic methods, and mechanistic interpretability. I am interested in the limitations of machine learning methods and how these methods scale.

Alongside my studies, I was a Research Fellow at Apart Research, where I worked on mechanistic interpretability. I have also been involved with SAIN Groningen, including its discussion groups. I have taught and supervised student projects at the University of Groningen and completed internships in data science and machine learning at Novigenix and Nuwa Innovation.

I expect to graduate around January 2027 and am looking for a PhD position.

Contacts: matthijs.vanderlende@gmail.com | Google Scholar | OpenReview | GitHub | LinkedIn

News

Aug, 2026	Aletheia's Quest: SAIN Groningen placed second in the black-box track of this LLM lie-detection competition and won the Scalability Award.
May, 2026	Winner, AAMAS NS-Gym Competition: our PPO-based submission ranked first in all four evaluation criteria for agents adapting to non-stationary environments.
Jan, 2026	Recognized as an Outstanding Reviewer for NLDL 2026.
Dec, 2025	Spotlight, NeurIPS 2025 Mechanistic Interpretability Workshop: our paper on interpretable feature steering for preference optimization was presented as a spotlight.
Jan, 2025	Published work on Gaussian-process function approximation for reinforcement learning at NLDL 2025.

Publications

NeurIPS MI

The Anatomy of Alignment: Decomposing Preference Optimization by Steering Sparse Features

Jeremias Ferrao, Matthijs van der Lende, Ilija Lichkovski, and Clement Neo

In NeurIPS Mechanistic Interpretability Workshop, 2025

Spotlight arXiv Bib HTML PDF Code

Spotlight presentation

@inproceedings{ferrao2025anatomy,
  title = {The Anatomy of Alignment: Decomposing Preference Optimization by Steering Sparse Features},
  author = {Ferrao, Jeremias and {van der Lende}, Matthijs and Lichkovski, Ilija and Neo, Clement},
  booktitle = {NeurIPS Mechanistic Interpretability Workshop},
  year = {2025},
}

NLDL

Interpretable Function Approximation with Gaussian Processes in Value-Based Model-Free Reinforcement Learning

Matthijs van der Lende, Matthia Sabatelli, and Juan Cardenas-Cartagena

In Proceedings of the 6th Northern Lights Deep Learning Conference (NLDL), 2025

Bib HTML PDF Code

@inproceedings{pmlr-v265-lende25a,
  title = {Interpretable Function Approximation with Gaussian Processes in Value-Based Model-Free Reinforcement Learning},
  author = {{van der Lende}, Matthijs and Sabatelli, Matthia and Cardenas-Cartagena, Juan},
  booktitle = {Proceedings of the 6th Northern Lights Deep Learning Conference (NLDL)},
  pages = {141--154},
  year = {2025},
  volume = {265},
  series = {Proceedings of Machine Learning Research},
  publisher = {PMLR},
  url = {https://proceedings.mlr.press/v265/lende25a.html},
}