Steinar Laenen

PhD Candidate Computer Science

University of Edinburgh

Biography

I am a third-year PhD student at the University of Edinburgh’s School of Informatics, where I am supervised by Dr. He Sun. My PhD research focuses on developing clustering algorithms for graphs, using techniques from spectral graph theory.

Before that, I was a research intern at Five, in their Oxford research group, where I worked on few-shot learning together with Dr. Luca Bertinetto

I received my MSc in Artificial Intelligence graduate (with distinction) from the University of Edinburgh. For my thesis I received the Joint AI MSc Dissertation Prize, awarded to 2 out of 200+ students. I did my bachelors (summa cum laude) in mathematics and computer science at Amsterdam University College.

Education

PhD in Computer Science, 2021 - present
University of Edinburgh
MSc in Artificial Intelligence, 2018 - 2019
University of Edinburgh
BSc in Liberal Arts and Science, Maths and Computer Science, 2015 - 2018
Amsterdam University College

Publications

Steinar Laenen, He Sun

May 2024 to appear at ICML 2024

Dynamic Spectral Clustering with Provable Approximation Guarantee

We introduce an algorithm that can dynamically maintain the clusters of a graph as new edges and vertices are added.

Steinar Laenen, Bogdan Manghiuc, He Sun

July 2023 at ICML 2023

Hierarchical Clustering: A Nearly-Optimal Construction for Well-Clustered Graphs

We present algorithms for constructing nearly-optimal hierarchical clustering trees with respect to Dasgupta`s cost function on well-clustered graphs.

Steinar Laenen

July 2023 at TAGL ML workshop at ICML 2023

One-Shot Neural Network Pruning via Spectral Graph Sparsification

We introduce a neural network pruning algorithm using spectral graph sparsification, empirically demonstrating its effectiveness in preserving performance at higher sparsity levels and providing theoretical analyses of local and global network connectivity.

Steinar Laenen, Luca Bertinetto

December 2021 At NeurIPS 2021 (also Meta-Learning Workshop NeurIPS 2020)

On Episodes, Prototypical Networks, and Few-Shot Learning

We develop simple theory and use experiments to show that the episodic training setup for Matching Networks and Prototypical Networks is inefficient.

Steinar Laenen, He Sun

December 2020 At NeurIPS 2020

Higher-Order Spectral Clustering of Directed Graphs

We show that a path-like cluster structure can be found in directed graphs, by performing spectral clustering on the first eigenvector of a newly defined Hermitian adjacency matrix.

Grimur Bjornsson, Gunnar Grimsson, Ari Sigurdsson, Steinar Laenen

February 2019 At 44th Workshop on Geothermal Reservoir Engineering

Thermal mapping of Icelandic geothermal surface manifestations with a drone

As one of the first to do this in the world, we use drones equipped with infrared and RGB cameras to map geothermal surface manifestations

Daniel Veutgen, Marco Massetti, Joy Rossi, Leonardo Veroli, Athena Asgeirsdottir, Gudrun Baldursdottir, Sigrun Gissurardottir, Gisli Gudmundsson, Tinna Sigurdardottir, Steinar Laenen, and others

November 2018 At 18th International Conference on Intelligent Virtual Agents

Interpreting Social Commitment in a Simulated Theater

Using theory from sociology, we develop bodily cues in virtual agents which effectively communicate social engagement.

Projects

Organisor of a talk series on societal impact of AI development

At Edinburgh University I was heavily involved in initiating and organising a series of panel discussions titled We Need to Talk About AI, where we invited insightful speakers to discuss some pressing societal issues around AI development. I was involved in three events which related to Data & Privacy, the Future of Work and Autonomous Weapons. Due to the succes of the talk series it is now officially funded and endorsed by the Edinburgh Futures Institute.

Research Team co-lead at EUFS

While doing my AI MSc at the University of Edinburgh, I was part of the autonomous driving team at Edinburgh University Forumula Student , where I co-led the research team. We implemented a deep ConvNet to drive the car autonomously in ROS simulation using supervised learning.

Experience

Research Intern

FiveAI

Dec 2019 – Dec 2020 Oxford, UK

Researched few-shot learning for image classification at FiveAI`s Oxford research group, which has close ties to the Torr Vision Group at Oxford University. The work I did together with Luca Bertinetto resulted in a contributed talk at the NeurIPS 2020 Meta Learning workshop, and a NeurIPS 2021 paper.

Summer Intern

Warm Arctic

Jul 2018 – Aug 2018 Reykjavik, Iceland

I worked on a project where we used drones equipped with RGB and infrared cameras to create orthomosaic maps of areas that contain geothermal lineaments. I developed an anomaly detection tool to automatically detect areas of geothermal importance using deep autoencoders and a one-class SVM. This work got published at the Stanford workshop on Geothermal Reservoir Engineering.

Software Engineering Intern

University of Reykjavik

Jul 2017 – Aug 2017 Reykjavik, Iceland

Worked on a project which aimed to research the importance of bodily cues in social interaction. The work consisted of designing and programming automated intelligent social behaviour for virtual agents in Unity3D and C#. Constructed complex social behaviours for virtual agents such as glancing, conversing, and gesticulating. This work got published at a virtual agents conference.