This year we received 8 nominations, which were all reviewed by several SAIS board members. In the end, the board was not able to distinguish between the two top theses, hence the award this year is shared by two winners. The official announcement is as follows:
The SAIS Best AI Master’s Thesis Award 2020 goes to Elias Lousseief, KTH, for the thesis entitled “MahlerNet: Unbounded Orchestral Music with Neural Networks”, and Ulme Wennberg, KTH, for the thesis entitled “An evaluation of BERT for a Span-based Approach for Jointly Predicting Entities, Coreference Clusters and Relations Between Entities”. Both the theses are extensive and well written, and treat problems at the core of AI research. Congratulations to the winners!
The thesis by Elias Lousseief treats the interesting open problem of music composition by machines, based on machine learning models. One of the strengths of the thesis is the extensive literature survey. Together with the clear and useful introduction to basic concepts this makes the thesis a valuable resource for anyone starting up work in the area. In addition, the thesis provides a novel solution for music generation, as well as valuable discussions on both technical and ethical aspects of the work, which has been published in a paper at the 2019 Nordic Sound and Music Computing conference. We look forward to an interesting presentation, perhaps including a musical performance, at SAIS workshop!
The thesis by Ulme Wennberg is in the area of Natural Language Processing, a core subfield of AI, and evaluates the use of recent language models for important NLP tasks. The evaluation is comprehensive and technically sound, with interesting results, and has been published as part of a paper at EMNLP2019. The work is a solid research contribution in the NLP area, as well as relevant for AI in general. The presentation of the thesis at SAIS workshop will attract a broad interest and open up for many interesting discussions!
Both the theses will be presented at the SAIS workshop 2020 (online), on June 16th.
Congratulations to the winners, and a big thanks to everyone who nominated a thesis!
The SAIS board,
through Eva Blomqvist, SAIS secretary
