AI Expo @ Skylab, DTU

Do you wonder if AI will take over the world? Learn more about the latest advancements in AI and see how you can contribute to shaping the future of humanity.

Come and mingle with fellow like-minded students and AI enthusiasts as we explore the latest state-of-the-art research in deep learning presented by several leading researchers in the field. Get a better understanding of how academic solutions are being used by the industry when KPMG presents their way of collaborating with the universities. Finally, you can end the evening enjoying pizza and beer, as we discuss our common interests in AI and beyond. It all starts at DTU Skylab!

Get you RSVP Today! Click on the ticket link. We will serve dinner and drinks.

17:00 to 17:30 Check-in
17:30 to 17:40: Opening words
17:40 to 18:30: From university to industry - KPMG
18:30 to 18:45: Deep learning topics by Nicki Detlefsen
18:45 to 19:00: Break and beers
19:00 to 19:20: Fair models from imbalanced data by Pola Schwobel
19:20 to 19:40: Cross-lingual word embeddings by Yova Kementchedjhieva
19:40 to late: Pizza and networking

Main sponsor: KPMG

Nicki Skafte Detlevsen

Bio: Nicki is in the third year of a PhD fellowship in Geometric Machine Learning under the supervision of Søren Hauberg and Ole Winter at Technical University of Denmark. His interest revolves around how to incorporate geometric transformations into deep learning architechtures to make them more robust and perform better, and related topics. Nicki has a MSc degree & BSc in matematical modelling, both from Technical University of Denmark.

Talk: Nicki will speak about uncertainty and variance estimation in deep learning models, and why this is important topic that needs further research.
In this talk, he will introduce some simple tricks to make variance estimation more robust and show results from a recent paper of his accept at this years Neurips conference.

Pola Schwöbel
https://www.linkedin.com/in/pola-schwoebel/

Bio: Pola is a PhD student in Machine Learning at DTU Compute, Section for Cognitive Systems. With supervisors Søren Hauberg and Kristoffer Hougaard Madsen, she works on Learned Data Augmentation for Deep Learning. Pola has a BSc in Mathematics from TU Berlin and a MSc in IT and Cognition from the University of Copenhagen.

Talk: Pola will speak about how the Data Augmentation techniques can be used to correct for biases in data sets, an important topic within the field of Algorithmic Fairness. Pola will give an overview of this quickly emerging research area and the connections to her own work.

Yova Radoslavova Kementchedjhieva
https://www.linkedin.com/in/yova-kementchedjhieva/

Bio: Yova is in the second year of a PhD fellowship in Natural Language Processing and Machine Learning under the supervision of Anders Søgaard and Isabelle Augenstein at the University of Copenhagen. Her interests revolve around word embeddings, with a focus on cross-lingual similarities and differences. Yova has a MSc degree in NLP and a BSc in Linguistics, both from the University of Edinburgh, Scotland.

Talk: Yova will speak on how Cross-lingual word embeddings are useful for the purposes of transfer learning and dictionary induction. In this talk, she will introduce a simple but very robust supervised method for linear alignment between independently learned word embeddings in different languages.

We are looking forward to an amazing evening!