177 citations · 186 across the 15 of their papers we have counts for
15 papers
Hierarchical Variational Policies for Reward-Guided Diffusion
Kushagra Pandey, Farrin Marouf Sofian, Jan Niklas Groeneveld +2
Adapting pretrained diffusion models to downstream objectives such as inverse problems often requires expensive test-time guidance or optimization. We propose a principled framewor…
Calibrated Test-Time Guidance for Bayesian Inference
Daniel Geyfman, Felix Draxler, Jan Groeneveld +3
Test-time guidance is a widely used mechanism for steering pretrained diffusion models toward outcomes specified by a reward function. Existing approaches, however, focus on maximi…
Parallel Token Prediction for Language Models
Felix Draxler, Justus Will, Farrin Marouf Sofian +3
Autoregressive decoding in language models is inherently slow, generating only one token per forward pass. We propose Parallel Token Prediction (PTP), a general-purpose framework f…
Beyond Accuracy: Are Time Series Foundation Models Well-Calibrated?
Coen Adler, Yuxin Chang, Felix Draxler +2
The recent development of foundation models for time series data has generated considerable interest in using such models across a variety of applications. Although foundation mode…
Variational Control for Guidance in Diffusion Models
Kushagra Pandey, Farrin Marouf Sofian, Felix Draxler +2
Diffusion models exhibit excellent sample quality, but existing guidance methods often require additional model training or are limited to specific tasks. We revisit guidance in di…
TRADE: Transfer of Distributions between External Conditions with Normalizing Flows
Stefan Wahl, Armand Rousselot, Felix Draxler +2
Modeling distributions that depend on external control parameters is a common scenario in diverse applications like molecular simulations, where system properties like temperature…
On the Universality of Volume-Preserving and Coupling-Based Normalizing Flows
Felix Draxler, Stefan Wahl, Christoph Schnörr +1
We present a novel theoretical framework for understanding the expressive power of normalizing flows. Despite their prevalence in scientific applications, a comprehensive understan…
Learning Distributions on Manifolds with Free-Form Flows
Peter Sorrenson, Felix Draxler, Armand Rousselot +2
We propose Manifold Free-Form Flows (M-FFF), a simple new generative model for data on manifolds. The existing approaches to learning a distribution on arbitrary manifolds are expe…
Free-form Flows: Make Any Architecture a Normalizing Flow
Felix Draxler, Peter Sorrenson, Lea Zimmermann +2
Normalizing Flows are generative models that directly maximize the likelihood. Previously, the design of normalizing flows was largely constrained by the need for analytical invert…
On the Convergence Rate of Gaussianization with Random Rotations
Felix Draxler, Lars Kühmichel, Armand Rousselot +3
Gaussianization is a simple generative model that can be trained without backpropagation. It has shown compelling performance on low dimensional data. As the dimension increases, h…
Lifting Architectural Constraints of Injective Flows
Peter Sorrenson, Felix Draxler, Armand Rousselot +3
Normalizing Flows explicitly maximize a full-dimensional likelihood on the training data. However, real data is typically only supported on a lower-dimensional manifold leading the…
Bose Einstein condensate as nonlinear block of a Machine Learning pipeline
Maurus Hans, Elinor Kath, Marius Sparn +5
Physical systems can be used as an information processing substrate and with that extend traditional computing architectures. For such an application the experimental platform must…
Finding Competence Regions in Domain Generalization
Jens Müller, Stefan T. Radev, Robert Schmier +3
We investigate a "learning to reject" framework to address the problem of silent failures in Domain Generalization (DG), where the test distribution differs from the training distr…
Whitening Convergence Rate of Coupling-based Normalizing Flows
Felix Draxler, Christoph Schnörr, Ullrich Köthe
Coupling-based normalizing flows (e.g. RealNVP) are a popular family of normalizing flow architectures that work surprisingly well in practice. This calls for theoretical understan…
Essentially No Barriers in Neural Network Energy Landscape
Felix Draxler, Kambis Veschgini, Manfred Salmhofer +1
Training neural networks involves finding minima of a high-dimensional non-convex loss function. Knowledge of the structure of this energy landscape is sparse. Relaxing from linear…