This is the personal web page of Mats Sjöberg. I live in Espoo, Finland together with my wife and son. I like reading, walking and programming. I care about social and economic justice.

I work as a Machine learning specialist at CSC. I received my doctoral degree in computer science from Aalto University in 2014, and I’m particularly interested in deep learning and computer vision.

Contact details

The best way to contact me is via email: mats@sjoberg.fi. I also use Signal for messaging.

Blog

This web page also contains my (rather infrequently updated) blog. The most recent posts are:

Surveillance capitalism
Posted 11 August 2019.
Simplifying my digital life
Posted 04 July 2018.

My blog also has an RSS feed. Add that link to a Feed reader to get updates whenever there is a new blog post.

Software

Here are links to some free and open source software I have created, some professionally and some in my free time.

DeepCaption is an image captioning framework based on deep learning. It uses the PyTorch platform, and was created while I was working at Aalto University.
Digital Me (DiMe) – the idea of DiMe was to collect your personal data from various loggers into a central place that you control. The focus is on tracking knowledge work and developing applications that can help the knowledge worker manage their working life better. Created while working at the University of Helsinki.
Pair annotate is a web-based tool for performing pair-wise annotation of multimedia (videos or images), used in several MediaEval tasks.
Pumpa is a simple pump.io client written in C++ and Qt. Pumpa was mentioned on LWN: “Searching for a pump.io client”.
Heebo was a simple but addictive game in the Match-3 genre for the Nokia N9 and newer Nokia Symbian. The game got over 100,000 downloads in the Nokia Ovi Store! It’s still available in the Opera Mobile Store. There is also a Jolla port.
Yaics is a simple GNU Social client in C++ and Qt. Since November 2014 Stig Atle Steffensen took over as maintainer of the project.
getpods is a trivial podcast client in Python.
ColourSOM is a simple HTML5 canvas + JavaScript application that demonstrates the Self-Organising Map algorithm by using random colours.
JSMandel just draws the Mandelbrot set.