Skip to main content
conference lunch move company map contacts lindholmen lindholmen 2 travel info

Decentralized AI – Two researchers to AI Sweden/KTH research project

We are now looking for 2 researchers to work with Decentralized AI. The project will be led by AI Sweden in collaboration with KTH. The ideal candidates have a PhD in machine learning, theoretical or applied, strong mathematical background, and excellent programming skills preferably using Python.

Man sketching on a whiteboard, visualization of decentralization of AI

The current state of the art for developing Artificial Intelligence (AI) relies on a centralized approach, where data is communicated to a central computer cluster(s) that learn models of a target application based on the data received. However, in many important applications, we have a system of different agents that possess local/private data sets. The amount of data per agent could be either too small or could be improved by learning from other agents’ data or other shared information between the agents. A centralized approach of training machine learning models may not be viable because of constraints in the system such as privacy, communication limitations and/or computational limitations. Therefore, a decentralized approach is needed here. While there are many interesting research avenues, AI Sweden has chosen to focus on the problem that we anticipate to most likely be the most common in applications, namely the problem where communicating any form of personal data is prohibited. This is also the hardest problem and is open in its full generality.


About AI Sweden

AI Sweden is the national center for applied artificial intelligence, supported by the Swedish government and the public and private sectors across the country. AI Sweden’s mission is to increase the use of AI for the benefit of our society, our competitiveness, and for everyone living in Sweden.

Application form

About you

Academic CV

Files must be less than 50 MB.
Allowed file types: pdf.

Transcript of records

Files must be less than 50 MB.
Allowed file types: pdf.