1 code implementation • SIGDIAL (ACL) 2021 • Markus Brenneis, Maike Behrendt, Stefan Harmeling
Exchanging arguments is an important part in communication, but we are often flooded with lots of arguments for different positions or are captured in filter bubbles.
no code implementations • 12 Sep 2024 • Maike Behrendt, Stefan Sylvius Wagner, Stefan Harmeling
Online spaces allow people to discuss important issues and make joint decisions, regardless of their location or time zone.
no code implementations • 18 Jun 2024 • Stefan Sylvius Wagner, Maike Behrendt, Marc Ziegele, Stefan Harmeling
In this work, we show how to leverage LLM-generated synthetic data to train and improve stance detection agents for online political discussions:(i) We generate synthetic data for specific debate questions by prompting a Mistral-7B model and show that fine-tuning with the generated synthetic data can substantially improve the performance of stance detection.
no code implementations • 11 Apr 2024 • Stefan Sylvius Wagner, Maike Behrendt, Marc Ziegele, Stefan Harmeling
In this work, we present two different ways to leverage LLM-generated synthetic data to train and improve stance detection agents for online political discussions: first, we show that augmenting a small fine-tuning dataset with synthetic data can improve the performance of the stance detection model.
1 code implementation • 3 Apr 2024 • Maike Behrendt, Stefan Sylvius Wagner, Marc Ziegele, Lena Wilms, Anke Stoll, Dominique Heinbach, Stefan Harmeling
In this work, we introduce AQuA, an additive score that calculates a unified deliberative quality score from multiple indices for each discussion post.
1 code implementation • 5 Feb 2024 • Stefan Sylvius Wagner, Stefan Harmeling
In this paper we adopt a representation-centric perspective on exploration in reinforcement learning, viewing exploration fundamentally as a density estimation problem.
1 code implementation • 8 Dec 2023 • Marc Höftmann, Jan Robine, Stefan Harmeling
Can we learn policies in reinforcement learning without rewards?
1 code implementation • 13 Sep 2023 • Thomas Germer, Jan Robine, Sebastian Konietzny, Stefan Harmeling, Tobias Uelwer
A CT scanner consists of an X-ray source that is spun around an object of interest.
no code implementations • 30 Aug 2023 • Stefan Sylvius Wagner, Peter Arndt, Jan Robine, Stefan Harmeling
In environments with sparse rewards, finding a good inductive bias for exploration is crucial to the agent's success.
2 code implementations • 22 Aug 2023 • Tobias Uelwer, Jan Robine, Stefan Sylvius Wagner, Marc Höftmann, Eric Upschulte, Sebastian Konietzny, Maike Behrendt, Stefan Harmeling
Learning meaningful representations is at the heart of many tasks in the field of modern machine learning.
1 code implementation • 13 Mar 2023 • Jan Robine, Marc Höftmann, Tobias Uelwer, Stefan Harmeling
Deep neural networks have been successful in many reinforcement learning settings.
Model-based Reinforcement Learning
reinforcement-learning
+2
1 code implementation • 13 Jan 2023 • Marc Höftmann, Jan Robine, Stefan Harmeling
Very large state spaces with a sparse reward signal are difficult to explore.
1 code implementation • NeurIPS CellSeg 2022 2022 • Eric Upschulte, Stefan Harmeling, Katrin Amunts, Timo Dickscheid
In the context of the NeurIPS 22 Cell Segmentation Challenge, the proposed solution is shown to generalize well in a multi-modality setting, while respecting domain-specific requirements such as focusing on specific cell types.
no code implementations • 26 Oct 2022 • Maximilian Kertel, Stefan Harmeling, Markus Pauly
Many production processes are characterized by numerous and complex cause-and-effect relationships.
no code implementations • 1 Oct 2022 • Leonid Kostrykin, Stefan Harmeling
In this paper, we consider multi-frame blind deconvolution (MFBD), where image blur is described by the convolution of an unobservable, undeteriorated image and an unknown filter, and the objective is to recover the undeteriorated image from a sequence of its blurry and noisy observations.
1 code implementation • 31 May 2022 • Tobias Uelwer, Sebastian Konietzny, Stefan Harmeling
With extensive experiments on the Fourier phase retrieval problem and thorough ablation studies, we can show the benefits of our modified ILO and the new initialization schemes.
1 code implementation • 30 May 2022 • Thomas Germer, Tobias Uelwer, Stefan Harmeling
Our method consists of three steps: First, we estimate a warping transformation of the images to align the sharp images with the blurred ones.
1 code implementation • NeurIPS Workshop Deep_Invers 2021 • Tobias Uelwer, Nick Rucks, Stefan Harmeling
In this work, we consider a modified version of the phase retrieval problem, which allows for a reference image to be added onto the image before the Fourier magnitudes are measured.
no code implementations • 18 Jun 2021 • Tobias Uelwer, Tobias Hoffmann, Stefan Harmeling
Fourier phase retrieval is the problem of reconstructing a signal given only the magnitude of its Fourier transformation.
no code implementations • 18 Jun 2021 • Stefan Wagner, Michael Janschek, Tobias Uelwer, Stefan Harmeling
We propose a new approach to increase inference performance in environments that require a specific sequence of actions in order to be solved.
2 code implementations • 7 Apr 2021 • Eric Upschulte, Stefan Harmeling, Katrin Amunts, Timo Dickscheid
We construct CPN models with different backbone networks, and apply them to instance segmentation of cells in datasets from different modalities.
no code implementations • 9 Mar 2021 • Christian Schiffer, Stefan Harmeling, Katrin Amunts, Timo Dickscheid
By solving the brain mapping problem on this graph using graph neural networks, we obtain significantly improved classification results.
no code implementations • 25 Nov 2020 • Christian Schiffer, Hannah Spitzer, Kai Kiwitz, Nina Unger, Konrad Wagstyl, Alan C. Evans, Stefan Harmeling, Katrin Amunts, Timo Dickscheid
Here we present a new workflow for mapping cytoarchitectonic areas in large series of cell-body stained histological sections of human postmortem brains.
no code implementations • 25 Nov 2020 • Christian Schiffer, Katrin Amunts, Stefan Harmeling, Timo Dickscheid
Cytoarchitectonic maps provide microstructural reference parcellations of the brain, describing its organization in terms of the spatial arrangement of neuronal cell bodies as measured from histological tissue sections.
no code implementations • 12 Oct 2020 • Jan Robine, Tobias Uelwer, Stefan Harmeling
Sample efficiency remains a fundamental issue of reinforcement learning.
Ranked #27 on
Atari Games
on Atari 2600 Pong
1 code implementation • 26 Jun 2020 • Thomas Germer, Tobias Uelwer, Stefan Conrad, Stefan Harmeling
Alpha matting aims to estimate the translucency of an object in a given image.
1 code implementation • 25 Mar 2020 • Thomas Germer, Tobias Uelwer, Stefan Conrad, Stefan Harmeling
Alpha matting describes the problem of separating the objects in the foreground from the background of an image given only a rough sketch.
1 code implementation • 10 Dec 2019 • Tobias Uelwer, Alexander Oberstraß, Stefan Harmeling
In this paper, we propose the application of conditional generative adversarial networks to solve various phase retrieval problems.
1 code implementation • 9 Jun 2019 • Felix Michels, Tobias Uelwer, Eric Upschulte, Stefan Harmeling
This paper extensively evaluates the vulnerability of capsule networks to different adversarial attacks.
1 code implementation • 5 Feb 2019 • Felix Dangel, Stefan Harmeling, Philipp Hennig
We propose a modular extension of backpropagation for the computation of block-diagonal approximations to various curvature matrices of the training objective (in particular, the Hessian, generalized Gauss-Newton, and positive-curvature Hessian).
no code implementations • 13 Jun 2018 • Hannah Spitzer, Kai Kiwitz, Katrin Amunts, Stefan Harmeling, Timo Dickscheid
We show that the self-supervised model has implicitly learned to distinguish several cortical brain areas -- a strong indicator that the proposed auxiliary task is appropriate for cytoarchitectonic mapping.
no code implementations • 30 May 2017 • Hannah Spitzer, Katrin Amunts, Stefan Harmeling, Timo Dickscheid
Its high resolution allows the study of laminar and columnar patterns of cell distributions, which build an important basis for the simulation of cortical areas and networks.
no code implementations • 28 Jun 2014 • Christian J. Schuler, Michael Hirsch, Stefan Harmeling, Bernhard Schölkopf
We describe a learning-based approach to blind image deconvolution.
no code implementations • IEEE Transactions on Pattern Analysis and Machine Intelligence 2013 • Christoph H. Lampert, Hannes Nickisch, Stefan Harmeling
To tackle the problem, we introduce attribute-based classification: Objects are identified based on a high-level description that is phrased in terms of semantic attributes, such as the object’s color or shape.
no code implementations • CVPR 2013 • Christian J. Schuler, Harold Christopher Burger, Stefan Harmeling, Bernhard Scholkopf
In this work, we also rely on a two-step procedure, but learn the second step on a large dataset of natural images, using a neural network.
no code implementations • CVPR 2013 • Stefan Harmeling, Michael Hirsch, Bernhard Scholkopf
We establish a link between Fourier optics and a recent construction from the machine learning community termed the kernel mean map.
no code implementations • NeurIPS 2010 • Stefan Harmeling, Hirsch Michael, Bernhard Schölkopf
Modelling camera shake as a space-invariant convolution simplifies the problem of removing camera shake, but often insufficiently models actual motion blur such as those due to camera rotation and movements outside the sensor plane or when objects in the scene have different distances to the camera.
no code implementations • 6 Dec 2009 • David Baehrens, Timon Schroeter, Stefan Harmeling, Motoaki Kawanabe, Katja Hansen, Klaus-Robert Mueller
After building a classifier with modern tools of machine learning we typically have a black box at hand that is able to predict well for unseen data.