DOI,IST REx ID,Research Group,Title of publication
10.52202/085713-3966,22827,"DaAl,GradSch",HALO: Hadamard-assisted lower-precision optimization for LLMs
10.52202/085713-3769,22831,FrLo,Connecting neural models latent geometries with relative geodesic representations
10.52202/085713-2479,21076,"FrLo,SyCr",Prediction-powered causal inferences
10.52202/085713-3752,21072,FrLo,Head pursuit: Probing attention specialization in multimodal transformers
10.52202/085713-1444,21068,FrLo,The third pillar of causal analysis? A measurement perspective on causal representations
10.52202/085713-3125,21070,FrLo,Out-of-Distribution detection with relative angles
10.52202/085713-4798,22823,ChLa,Setting ε is not the issue in differential privacy
10.52202/079017-4156,18890,"GradSch,MaMo",Average gradient outer product as a mechanism for deep neural collapse
10.52202/079017-0560,18875,"GradSch,ChLa",Banded square root matrix factorization for differentially private model training
10.52202/079017-1343,18996,FrLo,Identifying general mechanism shifts in linear causal representations
10.52202/079017-2290,19005,"CaMu,FrLo",Marrying causal representation learning with dynamical systems for science
10.52202/079017-0299,19512,MoHe,Continual counting with gradual privacy expiration
10.52202/079017-2115,19515,FrLo,Latent functional maps: A spectral framework for representation alignment
10.52202/079017-0820,19509,"SyCr,FrLo",Smoke and mirrors in causal downstream tasks
null,14921,"MaMo,ChLa",Deep neural collapse is provably optimal for the deep unconstrained features model
10.52202/075280-2050,14954,FrLo,Assumption violations in causal discovery and the robustness of score matching
null,15023,"ToHe,KrCh",Compositional policy learning in stochastic control systems with formal guarantees
10.52202/075280-0147,14953,FrLo,Sample complexity bounds for score-matching: Causal discovery and generative modeling
null,12537,MaMo,Memorization and optimization in deep neural networks with minimum over-parameterization
null,14326,FrLo,Object-centric learning with slot attention