2025 Workshop on Self-Supervised Learning for Signal Decoding

Description

To date, most successful applications of deep learning in signals and decoding are based on supervised learning. However, supervised learning is contingent on the availability of labeled data. The need for labeled data is a serious limitation to applications at scale and complicates the maintenance of real-life supervised learning systems. The typical situation is that unlabeled data is abundant, and this has given rise to paradigms such as self-supervised learning (SSL).

SSL combines vast amounts of unlabeled data with limited labeled data, relying on a supervisory signal derived directly from the data itself. While SSL has seen many successful applications, notably in large language models (LLMs), the theoretical understanding of its underlying mechanisms is only starting to emerge, and the huge potential of pre-trained models is yet to be fully explored.

In this workshop, we will discuss both the emerging theoretical understanding and practical applications of SSL in decoding of complex signals. The program features invited talks, posters, and social activities.

Program

August 14
11:00 –  12:00     Welcome & Registration
12:00 – 13:00     Lunch
13:00 – 13:05     Opening 
13:05 – 14:05     Sebastian Roland Lapuschkin, Fraunhofer – Heinrich Hertz Institute
                               Invited talk: Citable LLMs (tentative)
14:05 – 14:45     Lenka Tetkova, Technical University of Denmark
                               Invited talk: On convex decision regions in deep networks
14:45 – 15:45     Poster session & coffee break
15:45 – 16:45     Hung-yi Lee, National Taiwan University
                               Invited talk: Teaching LLM to listen and speak
16:45 – 17:00     Coffee break
17:00 – 17:40     Sarthak Yadav, Aalborg University
                              Invited talk: Beyond transformers: alternatives to self-attention for
                             learning self-supervised audio representations
17:40 – 18:20    Sneha Das, Technical University of Denmark
                              Invited talk: Voices in the margins: bias, fairness, and privacy in
                             speech processing systems
19:00 – 21:30   Banquet at Musikkens Hus, Musikkens Plads 1, 9000 Aalborg
 21:30                  End of the day

August 15
09:00 – 10:00   Jinyu Li, Microsoft
                              Invited talk: Unlocking multimodal intelligence with large language
                             model (IEEE SPS Distinguished Industry Speaker lecture)
10:00 – 10:30   Coffee break
10:30 – 11:10    Lars Kai Hansen, Technical University of Denmark
                              Invited talk: Self-supervised learning theory – old and new results
11:10 – 11:50      Zheng-Hua Tan, Aalborg University
                              Invited talk: Self-supervised learning for speech and audio applications
12:00 – 13:00    Lunch & goodbye 

Registration is free with lunches, dinner and coffee breaks included.

There is a no-show fee of DKK 2000.

Space is limited, so please sign up ASAP and no later than July 31, 2025. Cancellations are accepted no later than August 7, 2025.

Sign up below.