Poster Session GraML & ELLIS Unit Graz
On June 17, 2026, we participated in the ELLIS Sites ICML Fest 2026: Celebrate, Connect, Collaborate. Together with the Graz Center for Machine Learning (GraML) we hosted a Poster Session for the AI community at TU Graz to celebrate recent work, foster the scientific exchange within our network and strengthen our internal collaboration.
More than 20 poster contributions included work that was or will be presented at e.g., ICML 2026, ICLR 2026, AIROV 2026, RecSys 2025, and CPAL 2025. For a list of titles, see below.
A huge thank you to everyone who presented their work, asked thought-provoking questions, and contributed to making this event such a success. We are already looking forward to the next one!
Posters
| Authors | Title | Venue |
|---|---|---|
| Alexander Mayr, Robert Legenstein | Extending the Computational Capabilities of SSMs Using Recurrent Layers | - |
| Andre Koczka, Lukas Wimmer, Andre Koczka, Uros Petrovic, Gerald Steinbauer-Wagner | Overcoming Nature: Perception for Autonomous Navigation in Dense Vegetation | ICRA 2026 - Field Robotics Workshop |
| Benedikt Tscheschner, Eduardo Veas, Marc Masana | One Loop, Two Gains: Can Active Learning win the Lottery for Free? | Conference on Lifelong Learning Agents 2026 |
| Dong Wang | FOSL: A Foldable Sparse-and-Low-Rank Method for Efficient LLM Pre-training | ICLR and CPAL |
| Eduardo Veas, Ammaar Zaman, Oliver Prentner, Viksit Singh, Julian Kravanja, Laurenz Fussenegger, Patrick Hanfstringl | Embodied AI and Robot Learning | - |
| Florian Atzenhofer-Baumgartner, Georg Vogeler, Dominik Kowald | A Multistakeholder Approach to Value-Driven Co-Design of Recommender Systems Evaluation Metrics in Digital Archives | RecSys 2025 |
| Georg Niess, Roman Kern | SpanDiffusion: Flow Matching over Continuous Span Masks for Financial Causal Question Answering | Language Resources and Evaluation Conference 2026 |
| Gianluca Guglielmo, Marc Masana, Thomas Pock | Ranked Activation Shift for Post-Hoc Out-of-Distribution Detection | - |
| Hanne Say Eren, Ozan Özdeniczi | Object-Centric DLGNs for Abstract Visual Reasoning | - |
| Jeremy Chan Ho Tung, Eduardo Veas | Feature Importance Estimate via Embedded Neural Network Extension | IEEE Systems, Man and Cybernetics |
| Julian Gutheil, Simon Hitzginger, Robert Legenstein | Winner-Take-All bottlenecks enforce disentangled symbolic representations in multi-task learning | - |
| Kevin Innerebner, Markus Reiter-Haas, Elisabeth Lex | Neuro-symbolic Interpretable Recommender Systems | Human-Centred Computing Day, Austrian Computer Science Day |
| Maximilian Toller | Open Challenges in Machine Learning: Discontinuity, Scale Invariance, and Diverging Moments | - |
| Michael Obermayr, Robert Peharz | Joint Bayesian Inference on Lagrangian Physics and Trajectories | AIROV 2026 |
| Nikolaus Kopp, Fynn Gerding, Franz Pernkopf, Bernhard Geiger | Information-Theoretic Analysis of Neural Networks | - |
| O. Saukh, D. Wang, H. Šikić, Y. Cheng, L. Thiele | Cut Less, Fold More: Model Compression through the Lens of Projection Geometry | ICLR 2026 |
| Ozan Özdenizci | Privacy-Aware Lifelong Learning | ICLR 2025 |
| Ozan Özdenizci | Adversarially Robust Spiking Neural Networks with Sparse Connectivity | CPAL 2025 |
| Patrick Krumpl, Francesco Corti, Dong Wang, Olga Saukh | Deploying TinyML on Commercial Smart Glasses | To be submitted |
| Paul Horvath, Marian Staggl, Stefan Posch | Derivative-Enhanced Training for Data-efficient Surrogate Modeling | AIROV 2026 |
| Wenwu Tang | GRAIL: Post-hoc Compensation by Linear Reconstruction for Compressed Networks | Conference on Parsimony and Learning (CPAL) |
| Yeganeh Bahariasl, Simon Hitzginger, Robert Legenstein | Scalable semi-online training of recurrent spiking and artificial neural networks with chunk-wise truncated backpropagation | ICNCE 2026 |
Pictures
© Goertz
