Pre-NeurIPS ELLIS Fest 2024

Join us at the Institute of Science and Technology Austria (ISTA) for an exciting symposium to showcase ISTA's recent publications that will be presented at the Neural Information Processing Systems (NeurIPS), the flagship AI conference, taking place December 10-15 in Vancouver, Canada. 

This event will consist of a poster session with a selection of the papers co-authored by ISTA researchers and a series of talks, showcasing some of the work ongoing in the unit. 

At NeurIPS, ISTA researchers from the Vienna ELLIS unit will present 15 papers: 

1. "MicroAdam: Accurate Adaptive Optimization with Low Space Overhead and Provable Convergence" Ionut-Vlad Modoranu, Mher Safaryan, Grigory Malinovsky, Eldar Kurtić, Thomas Robert, Peter Richtarik, Dan Alistarh

2. "PV-Tuning: Beyond Straight-Through Estimation for Extreme LLM Compression" Vladimir Malinovskii, Denis Mazur, Ivan Ilin, Denis Kuznedelev, Konstantin Burlachenko, Kai Yi, Dan Alistarh, Peter Richtarik

3. "QuaRot: Outlier-Free 4-Bit Inference in Rotated LLMs" Saleh Ashkboos, Amirkeivan Mohtashami, Maximilian Croci, Bo Li, Pashmina Cameron, Martin Jaggi, Dan Alistarh, Torsten Hoefler, James Hensman

4. "The Iterative Optimal Brain Surgeon: Faster Sparse Recovery by Leveraging Second-Order Information" Diyuan Wu, Ionut-Vlad Modoranu, Mher Safaryan, Denis Kuznedelev, Dan Alistarh

5. "Banded Square Root Matrix Factorization for Differentially Private Model Training" Kalinin Nikita, Christoph Lampert

6. "Neural collapse vs. low-rank bias: Is deep neural collapse really optimal?" Peter Súkeník, Christoph Lampert, Marco Mondelli

7. "Matrix Denoising with Doubly Heteroscedastic Noise: Fundamental Limits and Optimal Spectral Methods" Yihan Zhang, Marco Mondelli

8. "Average gradient outer product as a mechanism for deep neural collapse" Daniel Beaglehole, Peter Súkeník, Marco Mondelli, Misha Belkin

9. "Continual Counting with Gradual Privacy Expiration" Joel Daniel Andersson, Monika Henzinger, Rasmus Pagh, Teresa Anna Steiner, Jalaj Upadhyay

10. "Identifying General Mechanism Shifts in Linear Causal Representations" Tianyu Chen, Kevin Bello, Francesco Locatello, Bryon Aragam, Pradeep Ravikumar

11. "Identifiable Object-Centric Representation Learning via Probabilistic Slot Attention" Avinash Kori, Francesco Locatello, Ainkaran Santhirasekaram, Francesca Toni, Ben Glocker, Fabio De Sousa Ribeiro

12. "Latent Functional Maps: a spectral framework for representation alignment" Marco Fumero, Marco Pegoraro, Valentino Maiorca, Francesco Locatello, Emanuele Rodolà

13. "Smoke and Mirrors in Causal Downstream Tasks" Riccardo Cadei, Lukas Lindorfer, Sylvia Cremer, Cordelia Schmid, Francesco Locatello

14. "Marrying Causal Representation Learning with Dynamical Systems for Science" Dingling Yao, Caroline Muller, Francesco Locatello

15. "$C^2M^3$: Cycle-Consistent Multi-Model Merging" Donato Crisostomi, Marco Fumero, Daniele Baieri, Florian Bernard, Emanuele Rodolà

The Institute of Science and Technology Austria (ISTA) is home to the ELLIS Unit Vienna, a locally organized unit of the European Laboratory for Learning and Intelligent Systems (ELLIS). As a member of this continent-wide artificial intelligence (AI) research excellence network, the unit distinguishes itself by its broad spectrum and interdisciplinarity. Embedded into the ISTA ecosystem, the unit focuses on excellence in advancing machine learning (ML) foundations and applications in science.