Page Not Found
Page not found. Your pixels are in another canvas.
Sitemap
A list of all the posts and pages found on the site. For you robots out there is an XML version available for digesting as well.
Pages
About me
About me
Page not in menu
This is a page not in th emain menu
Posts
coding
FTPnet
Published:
Implementation of a FTP client in .NET 3.5 for desktop and portable applications.
Volume Mixer Plus
Published:
Improves the default Windows Volume Mixer to allows one to change the default playback device by using shortcuts.
NNCreator
Published:
GUI to create network graphs for the CNTK framework, which afterwards can be trained and evaluated in C++.
Data Viewer
Published:
Utility tool to inspect various data formats to verify data integrity in machine learning.
Keras Utility & Layer Collection
Published:
Collection of custom layers for Keras which are missing in the main framework. These layers might be useful to reproduce current state-of-the-art deep learning papers using Keras.
Show, Attend and Tell for Keras
Published:
Keras implementation of the paper Show, Attend and Tell.
Transformer: Attention is all you need for Keras
Published:
Implementation of the Transformer architecture described by Vaswani et al. in Attention Is All You Need.
Lidar Viewer
Published:
Point cloud viewer with surface reconstruction for LIDAR data using OpenGL.
EmoMatch
Published:
Unsupervised Audio + Video Network Pretraining using PyTorch based on the correlation between audio and video signal.
VGGVox for PyTorch
Published:
Implementation of the VGGVox network in PyTorch.
Bayesian Spiking Neural Networks
Published:
Implementation of the paper Homeostatic plasticity in Bayesian spiking networks as Expectation Maximization with posterior constraints by Habenschuss et al.
Faster Training of Mask R-CNN by Focusing on Instance Boundaries
Published:
Reference implementation of Faster Training of Mask R-CNN by Focusing on Instance Boundaries.
portfolio
Portfolio item number 1
Short description of portfolio item number 1
Portfolio item number 2
Short description of portfolio item number 2
publications
Prediction spatio-temporal dynamics using reservoir computing
Published in University of Göttingen, 2017
Examination of recurrent neural networks (Echo State Networks) for the multiple prediction tasks in chaotic systems.
Zimmermann, Roland S. (2017). Prediction spatio-temporal dynamics using reservoir computing.
https://github.com/FlashTek/rcp_spatio_temporal/raw/master/paper/thesis/latex/thesis.pdfFaster Training of Mask R-CNN by Focusing on Instance Boundaries
Published in arXiv, 2018
Improving the training of Mask R-CNN for instance segmentation by introducing an intuitive auxiliary loss.
Zimmermann, R. S. and Siems, J. N. (2018). Faster Training of Mask R-CNN by Focusing on Instance Boundaries. arXiv preprint arXiv:1809.07069.
https://arxiv.org/abs/1809.07069Observing spatio-temporal dynamics of excitable media using reservoir computing
Published in Chaos: An Interdisciplinary Journal of Nonlinear Science, 2018
Examination of recurrent neural networks (Echo State Networks) for the spatio-temporal cross prediction of chaotic systems.
Zimmermann, R. S. and Parlitz, U. (2018). Observing spatio-temporal dynamics of excitable media using reservoir computing. Chaos: An Interdisciplinary Journal of Nonlinear Science, 28(4), 043118.
https://aip.scitation.org/doi/abs/10.1063/1.5022276Predicting and Observing Chaotic Dynamics in Excitable Media Using Machine Learning
Published in CinC, 2018
Analysis of chaotic dynamics using sequential and recurrent neural networks and classical Machine Learning methods.
Parlitz, U. and Zimmermann, R. S. , Herzog, S. and Isenseee, J. and Datseris, G. (2018). Predicting and Observing Chaotic Dynamics in Excitable Media Using Machine Learning. CinC 2018, Maastrich.
Comment on “Adv-BNN: Improved Adversarial Defense through Robust Bayesian Neural Network”
Published in arXiv, 2019
Breaks the the proposed adversarial defense method Adv-BNN
Zimmermann, R. S. (2019). Comment on “Adv-BNN: Improved Adversarial Defense through Robust Bayesian Neural Network”. arXiv preprint arXiv:1907.00895.
https://arxiv.org/abs/1907.00895Robust Perception in Humans and Machines
Published in University of Göttingen/University of Tübingen, 2020
Analyzing the robustness of computer vision and comparing it to human performance by developing novel evaluation methods
Zimmermann, Roland S. (2020). Robust Perception in Humans and Machines.
https://rzimmermann.com/files/Masters_Thesis_Roland_S_Zimmermann.pdfIncreasing the robustness of DNNs against image corruptions by playing the Game of Noise
Published in Towards Trustworthy ML: Rethinking Security and Privacy for ML (ICLR 2020 Workshop), 2020
Simple but properly tuned training with additive noise generalizes surprisingly well to unseen corruptions.
Rusak, E., Schott, L., Zimmermann, R.S., Bitterwolf, J., Bringmann, O., Bethge, M. and Brendel, W.,Increasing the robustness of DNNs against image corruptions by playing the Game of Noise.
https://trustworthyiclr20.github.io/rusak.pdfA simple way to make neural networks robust against diverse image corruptions
Published in European Conference on Computer Vision (ECCV) 2020, 2020
Simple but properly tuned training with additive noise generalizes surprisingly well to unseen corruptions.
Rusak, E., Schott, L., Zimmermann, R.S., Bitterwolf, J., Bringmann, O., Bethge, M. and Brendel, W.,A simple way to make neural networks robust against diverse image corruptions.
https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123480052.pdfFoolbox Native: Fast adversarial attacks to benchmark the robustness of machine learning models in PyTorch, TensorFlow, and JAX
Published in Journal of Open Source Software, 2020
Foolbox is a popular Python library to benchmark the robustness of machine learning models against these adversarial perturbations
Jonas, R. Zimmermann, R. S., Bethge, M. and Brendel, W., Foolbox Native: Fast adversarial attacks to benchmark the robustness of machine learning models in PyTorch, TensorFlow, and JAX.
https://joss.theoj.org/papers/10.21105/joss.02607.pdfNatural images are more informative for interpreting CNN activations than synthetic feature visualizations
Published in NeurIPS 2020 Workshop: Shared Visual Representations in Human & Machine Intelligence, 2020
Using human psychophysical experiments, we show that natural images can be significantly more informative for interpreting neural network activations than synthetic feature visualizations.
Borowski, J., Zimmermann, R. S., Schepers, J., Geirhos, R. , Wallis, T. S. A., Bethge, M. and Brendel, W., Natural images are more informative for interpreting CNN activations than synthetic feature visualizations .
https://openreview.net/forum?id=-vhO2VPjbVaContrastive Learning can Identify the Underlying Generative Factors of the Data
Published in NeurIPS 2020 Workshop: Self-Supervised Learning - Theory and Practice, 2020
We rigorously show that models trained on a common contrastive loss can implicitly invert the underlying generative model of the observed data up to affine transformations.
Zimmermann, R. S., Schneider, S. Sharma, Y., Bethge, M. and Brendel, W., Contrastive Learning can Identify the Underlying Generative Factors of the Data.
https://sslneuips20.github.io/files/CameraReadys%203-77/67/CameraReady/Contrastive_Learning_can_Identify_the_Underlying_Generative_Factors_of_the_Data_SSL_NeurIPS_2020.pdfContrastive Learning Inverts the Data Generating Process
Published in ICML 2021, 2020
We rigorously show that models trained on a common contrastive loss can implicitly invert the underlying generative model of the observed data up to affine transformations.
Zimmermann, R. S., Sharma, Y., Schneider, S. Bethge, M. and Brendel, W., Contrastive Learning Inverts the Data Generating Process.
https://sslneuips20.github.io/files/CameraReadys%203-77/67/CameraReady/Contrastive_Learning_can_Identify_the_Underlying_Generative_Factors_of_the_Data_SSL_NeurIPS_2020.pdfExemplary Natural Images Explain CNN Activations Better than State-of-the-Art Feature Visualization
Published in NeurIPS 2020 Workshop: Shared Visual Representations in Human & Machine Intelligence, 2021
Using human psychophysical experiments, we show that natural images can be significantly more informative for interpreting neural network activations than synthetic feature visualizations.
Borowski, J., Zimmermann, R. S., Schepers, J., Geirhos, R. , Wallis, T. S. A., Bethge, M. and Brendel, W., Natural images are more informative for interpreting CNN activations than synthetic feature visualizations .
https://openreview.net/forum?id=-vhO2VPjbVaReconstructing complex cardiac excitation waves from incomplete data using echo state networks and convolutional autoencoders
Published in Frontiers in Applied Mathematics and Statistics, 2021
We use two approaches from machine learning, echo state networks and convolutional autoencoders, to solve two relevant data modelling tasks in cardiac dynamics to bridge the gap between measurable and not measurable quantities
Herzog, S., Zimmermann, R. S., Abele, J., Luther, S. and Parlitz, U. (2021). Reconstructing Complex Cardiac Excitation Waves From Incomplete Data Using Echo State Networks and Convolutional Autoencoders. Front. Appl. Math. Stat. 6:616584. doi: 10.3389/fams.2020.616584
https://www.frontiersin.org/articles/10.3389/fams.2020.616584/fullScore-Based Generative Classifiers
Published in arXiv, 2021
We evaluate score-based generative models as classifiers on CIFAR-10 and find that they yield good accuracy and likelihoods but no adversarial robustness.
Zimmermann, R. S., Schott, L., Song, Y. Dunn, B. A. and Klindt, D. A. Score-Based Generative Classifiers
https://arxiv.org/abs/2110.00473How Well do Feature Visualizations Support Causal Understanding of CNN Activations?
Published in NeurIPS 2021, 2021
Using psychophysical experiments, we show that state-of-the-art synthetic feature visualizations do not support causal understanding much better than no visualizations, and only similarly well as other visualizations like natural dataset samples.
Zimmermann, R. S., Borowski, J., Geirhos, R. , Bethge, M., Wallis, T. S. A. and Brendel, W., How Well do Feature Visualizations Support Causal Understanding of CNN Activations?
https://openreview.net/forum?id=vLPqnPf9k0Increasing Confidence in Adversarial Robustness Evaluations
Published in NeurIPS 2022, 2022
We propose a test that enables researchers to find flawed adversarial robustness evaluations. Passing our test produces compelling evidence that the attacks used have sufficient power to evaluate the model’s robustness.
Zimmermann, R. S., Brendel, W., Tramer, F., and Carlini, N., Increasing Confidence in Adversarial Robustness Evaluations. NeurIPS 2022.
https://openreview.net/forum?id=NkK4i91VWpProvably Learning Object-Centric Representations
Published in ICML 2023, 2023
We analyze when object-centric representations can be learned without supervision and introduces two assumptions, compositionality and irreducibility, to prove that ground-truth object representations can be identified.
Brady, J., Zimmermann, R. S., Sharma, Y., Schölkopf, B., von Kügelken, J. and Brendel, W., Provably Learning Object-Centric Representations.
https://arxiv.org/pdf/2305.14229.pdfSensitivity of Slot-Based Object-Centric Models to their Number of Slots
Published in arXiv, 2023
We investigate the sensitivity of slot-based object-centric models to their number of slots, identify failures and explore mitigation strategies.
Zimmermann, R. S., van Steenkiste, S., Sajjadi, M. S. M., Kipf, T., Greff, K., Provably Learning Objet-Centric Representations.
https://arxiv.org/pdf/2305.18890.pdfDon’t trust your eyes: on the (un)reliability of feature visualizations
Published in arXiv, 2023
Feature visualizations seek to explain how neural networks process natural images, but as we show both experimentally and analytically, they can be unreliable (for instance, they can be manipulated to show arbitrary patterns).
Geirhos, R., Zimmermann, R. S., Bilodeau, B., Brendel, W., Kim, B., Don’t trust your eyes: on the (un)reliability of feature visualizations.
https://arxiv.org/pdf/2306.04719.pdfScale Alone Does not Improve Mechanistic Interpretability in Vision Models
Published in arXiv, 2023
We compare the mechanistic interpretability of vision models differing with respect to scale, architecture, training paradigm and dataset size and find that none of these design choices have any significant effect on the interpretability of individual units. We release a dataset of unit-wise interpretability scores that enables research on automated alignment.
Zimmermann, R. S., Klein, T. and Brendel, W., Scale Alone Does not Improve Mechanistic Interpretability in Vision Models.
https://arxiv.org/abs/2307.05471talks
Talk 1 on Relevant Topic in Your Field
Published:
This is a description of your talk, which is a markdown files that can be all markdown-ified like any other post. Yay markdown!
Conference Proceeding talk 3 on Relevant Topic in Your Field
Published:
This is a description of your conference proceedings talk, note the different field in type. You can put anything in this field.