Kategorien
Artikel

Deep Learning Summer School 2021 #DL4LASS

In a few minutes, the Summer School Deep Learning for Language Analysis (#DL4LASS) 2021 will come to an end with our panel session. Unfortunately, I didn’t had the time to write a detailed report this year, but since we largely stuck to last year’s plan, the diary from 2020 also provides quite good information about what we have organized.

Of course, organizing the school like in 2020 also meant that we had to rely on a virtual event again. Even though we succeeded quite well (at least judging by what I’ve heard), we would really like to be able to welcome the participants back to Cologne for next year’s edition. In fact, a virtual event causes much less work in a new edition because you can fall back on tried and tested procedures. Therefore, we have oriented ourselves relatively strongly on last year’s DL4LASS.

However, some things were new this year: While Nils Reiter from our Department (IDH) again organized the course on text data, this year we were able to win Paul Walbott from Fraunhofer IAIS for the audio workshop. This year we were able to win Thomas Oehlke and Moritz Eßer as tutors. There was also an important change in the technical infrastructure because we had access to two powerful NVIDIA DGX-1 machines with 8 GPUs each that the Center of Data and Simulation Science (CDS) purchased for the University of Cologne. Last year, we still had to rent machines from Amazon Web Service, which consumed the largest item in the budget of our Summer School. Thanks to Nils‘ configuration, the technology worked flawlessly and I think it will give us pleasure for the times to come.

Our panel discussion will be about perspectives on applications of Deep Learning methods in various areas of language processing. As speakers we invited Ekaterina Kruchinina (Cerence) who will talk about „Cerence NLU Research“, Timothée Schmude (CCeH/IDH) who discusses „Machine Learning in Criminal Law“ and Nasrin Saef (frogs42) with a contribution on „How to Train Your Language Model (so it doesn’t swear all the time)“. All three panel participants are alumni of our Master’s programme in Information Processing from different years. This again shows the different directions our graduates can take after their studies.

We are looking forward to this final session of our Summer School. If you want to participate in the panel at short notice, please send me an email or a message on Twitter. Or maybe we’ll see you at the next edition of our School in 2022. Further information will follow.


Von Jürgen Hermes

Arbeitet als Wissenschaftler und Geschäftsführer am Institut für Digital Humanities an der Universität zu Köln.

Schreibe einen Kommentar

Deine E-Mail-Adresse wird nicht veröffentlicht. Erforderliche Felder sind mit * markiert.

Diese Website verwendet Akismet, um Spam zu reduzieren. Erfahre mehr darüber, wie deine Kommentardaten verarbeitet werden.