A new wave of foundation models based on xLSTM

Wednesday 17 Dec 2025, 13:45 → 16:45 Europe/Vienna

Seminarraum 1/2 (PSK 3rd floor)

MLA²S

Topical Seminal for everyone working with timeseries data! The team of the JKU Linz that developed the "Extended Long Short-Term Memory xLSTM" will present us their latest results. 

There will also be time for discussions and networking!

 

Please register so we can estimate the catering. 

Zoom connection: [TBA shortly before meeting]

13:45 → 14:00 Arriving, grabbing a coffee, Welcome!

Speakers: Dr Claudius Krause (MBI Vienna (ÖAW)), Claus Trost (Erich Schmid Institute of Materials Science of theAustrian Academy of Sciences), Jan Odstrčilík (Institut für Mittelalterforschung, ÖAW), Kati Heinrich (IGF | ÖAW), Nicki Holighaus (Acoustics Research Institute, Austrian Academy of Sciences)

14:00 → 15:00 A new wave of foundation models based on xLSTM

xLSTM has emerged as powerful alternative to the Transformer-based architectures, on which most LLMs are built. The linear dependency on context size makes it suitable for many applications which require efficient processing of contextual data. This talk gives an overview of the development of recurrent neural networks, to Transformers, to linearized models, and their applications in application domains and science.

Speaker: Günter Klambauer (JKU Linz)

15:00 → 15:15 Discussion

15:15 → 16:15 Networking (with snacks)