Tatu Ylonen - Research Home Page

I will be updating this over the next few weeks as I find the time. Even adding some graphics, maybe!

This will be the personal and research home page of Tatu Ylonen (also known as Tatu Ylönen or Tatu Yloenen).

Computational Linguistics

I'm currently working on some new approaches to morphology and parsing for the Finnish language. There is not much public information on this yet, but check out the following projects on github:

I'm also working on various machine-learning based approaches to morphological learning, grammar learning, and knowledge extraction.

In the medium term I'm particularly interested in semantic parsing - extracting and representing the meaning of sentences.

Artificial Intelligence

I've always been interested in artificial intelligence, particularly language understanding, knowledge representation, resoning, and control.

I'm currently exploring various machine learning and knowledge representation approaches, especially in the areas of lexical semantics and common sense knowledge. While I'm fully aware of the wide popularity and success of deep learning in both vision and linguistics, my current approach is more towards combining traditional approaches with learning-based appoaches. In some cases these may involve gradient-based learning in a non-neural setting. In other cases my approach is more probability based. This work is currently still mostly unpublished (but see the patents page for some earlier work).

Cybersecurity

I wrote the original SSH (Secure Shell), which has since evolved into OpenSSH and various other versions. It now ships with every Unix, Linux, and Mac computer and is available for every platform. It is the de facto tool for system and network administration over computer networks. I've also done a lot of work in SSH key management, for both user authentication keys and host keys.

I'm currently interested in post-quantum cryptography and how to design security protocols for the post-quantum age. It does look like quantum computers might actually be becoming a reality. It is time to start preparing, as the changes are going to affect an awfully wide variety of protocols and systems, most of which will need to be upgraded or replaced - and many of those would normally only be replaced after 10-20 years!

Publications

Other links