• Home
• Publications
• Talks
• Curriculum Vitae
• GitHub

Hey!

I am Kokos [he/him]. I like to think, read, and talk about stuff at the intersection of formal logics, linguistics, and machine learning.

Whereabouts / Short CV

On my day to day, I am a post-doctoral researcher at Aalto University, where I do neurosymbolic reinforcement learning. Exceptionally, on Mondays and Tuesdays, I do climate stuff with neural networks for Skyfora. Beginning in September 2024, I will be visiting in the University of Bologna for a few months.

Before moving to Helsinki, I was located in Utrecht. I received my PhD (CompLing) and my master's degree (AI) from Utrecht University. While in Utrecht, I was also working as a machine learning engineer at geekbot. Before all that, I received my diploma in Electrical & Computer Engineering from the Aristotle University of Thessaloniki.

A longer version of the above can be found at my CV.

Thesis

My dissertation is titled Dependency as Modality, Parsing as Permutation: A Neurosymbolic Perspective on Categorial Grammars. It advocates the use of modalities as dependency annotations in type-logical grammars, and explores the parsing-as-deduction paradigm from a contemporary neurosymbolic perspective.

People thought it was good: it is one of the two recipients of the 2024 E. W. Beth Dissertation Prize. You can find a faithful copy of the physical print here. If (for whatever reason) you're interested in an actual physical copy, feel free to send me a message.

Research

My published research revolves around structured machine learning, neurosymbolic reasoning, and type-driven approaches to syntactic parsing and semantic compositionality, among other things. I am especially interested in the interface between formal and data-driven methods; if that floats your boat rubber duck, you can have a look at my publications page.

My latest paper is "On Tables with Numbers, with Numbers" (Preprint, August 2024).

Recent & Upcoming Talks

Here's a list of my recent and upcoming talks (more here):

• Dependency as Modality, Parsing as Permutation
  August 2, 2024 in Leuven
  Talk @ European Summer School of Logic, Language and Information (ESSLLI 2024)
• Geometry-Aware Supertagging with Heterogeneous Dynamic Convolutions
  September 12, 2023 in Göteborg
  Presentation @ Learning with Small Data
• Improving BERT Pretraining with Syntactic Supervision
  September 11, 2023 in Göteborg
  Poster @ Learning with Small Data
• SPINDLE: Spinning Raw Text into Lambda Terms with Graph Attention
  May 3, 2023 in Dubrovnik
  Poster @ The 17th Conference of the European Chapter of the Association for Computational Linguistics

Teaching

To my great displeasure, I am currently not teaching. Previously, I participated in various capacities in the following courses:

• Compositional Models of Vector-Based Semantics (ESSLLI 2022)
• Logic & Language (2020-2021, main lecturer)
• Logic & Language (2018-2019, 2019-2020, 2021-2022, lectures & project supervision)
• Machine Learning for Human Vision & Language (2019-2020, tutorials)
• Advanced Topics in Cognitive Modelling (2018-2019, tutorials)