Stefan Harmeling

I am the Group Leader of the Computational Imaging Group at the Max Planck Institute for Intelligent Systems (formerly Biological Cybernetics) in Prof Bernhard Schölkopf's department of Empirical Inference.

Before that I studied mathematics and logic at Universität Münster and computer science with a specialization on artificial intelligence at Stanford University. During my PhD I worked at the Fraunhofer Institut FIRST in Berlin where I got my PhD (Dr rer nat) from Universität Potsdam. While being a Marie-Curie fellow I spent two years at the Institute for Adaptive and Neural Computation at the University of Edinburgh.

email / biography / google scholar / arXiv / address

Stefan Harmeling


News



Research overview


Computational photography Machine learning Applications Miscellany

Publications


Multi-frame blind deconvolution and super-resolution

Online Multi-frame Blind Deconvolution with Super-resolution and Saturation Correction
M. Hirsch, S. Harmeling, S. Sra, B. Schölkopf
Astronomy and Astrophysics, 531 (A9), 2011

website and data sets

Multiframe Blind Deconvolution, Super-Resolution, and Saturation Correction via Incremental EM
S. Harmeling, M. Hirsch, S. Sra, B. Schölkopf
IEEE International Conference on Image Processing (ICIP), Hong Kong, 2010

Efficient Filter Flow for space-variant multiframe blind deconvolution
M. Hirsch, S. Sra, B. Schölkopf, S. Harmeling
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), San Francisco, USA, 2010

Online Blind Image Deconvolution for Astronomy
S. Harmeling, M. Hirsch, S. Sra, B. Schölkopf
IEEE Conference on Computational Photography (ICCP), San Francisco, USA, 2009

code (mbd-0.0.tar)



Camera shake removal

Recording and playback of camera shake: benchmarking blind deconvolution with a real-world database
R. Köhler, M. Hirsch, B. Schölkopf, B. Mohler, S. Harmeling
European Conference on Computer Vision (ECCV), Firenze, Italy, 2012

website and benchmark data sets

Fast Removal of Non-uniform Camera Shake
M. Hirsch, C. Schuler, S. Harmeling, B. Schölkopf
International Conference on Computer Vision (ICCV), Barcelona, Spain, 2011

website and images

Space-Variant Single-Image Blind Deconvolution for Removing Camera Shake
S. Harmeling, M. Hirsch, B. Schölkopf
Conference on Neural Information Processing Systems (NIPS), Volume 22, Vancouver, Canada, 2010




Optical aberration removal

Blind Correction of Optical Aberrations
C. Schuler, M. Hirsch, S. Harmeling, B. Schölkopf
European Conference on Computer Vision (ECCV), Firenze, Italy, 2012

website and images

Non-stationary Correction of Optical Aberrations
C. Schuler, M. Hirsch, S. Harmeling, B. Schölkopf
International Conference on Computer Vision (ICCV), Barcelona, Spain, 2011

website and images



Image deconvolution: miscellany

On a link between kernel mean maps and Fraunhofer diffraction, with an application to super-resolution beyond the diffraction limit
S. Harmeling, M. Hirsch, B. Schölkopf
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Portland, USA, 2013

A machine learning approach for image deconvolution
C. Schuler, H.C. Burger, S. Harmeling, B. Schölkopf
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Portland, USA, 2013

website and toolbox

Improving alpha matting and motion blurred foreground estimation
R. Köhler, M. Hirsch, B. Schölkopf, S. Harmeling
IEEE Conference on Image Processing (ICIP), Melbourne, Australia, 2013

Automatic Foreground Background Refocusing
A. Loktyushin, S. Harmeling
IEEE Conference on Image Processing (ICIP), Brussels, Belgium, 2011




Image denoising

Learning how to combine internal and external denoising methods
C. Burger, C. Schuler, S. Harmeling
Accepted at German Conference on Pattern Recognition (GCPR/DAGM), 2013

Image denoising with multi-layer perceptrons, part 1: comparison with existing algorithms and with bounds
C. Burger, C. Schuler, S. Harmeling
arXiv:1211.1544 [cs.CV], 2012

Image denoising with multi-layer perceptrons, part 2: training trade-offs and analysis of their mechanisms
C. Burger, C. Schuler, S. Harmeling
arXiv:1211.1544 [cs.CV], 2012

Image denoising: Can plain Neural Networks compete with BM3D?
H.C. Burger, C. Schuler, S. Harmeling
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Providence, USA, 2012

website with code and data

Improving Denoising Algorithms via a Multi-Scale Meta-Procedure
C. Burger, S. Harmeling
Annual Symposium of German Association for Pattern Recognition (DAGM), Frankfurt, Germany 2011 (DAGM prize)

Removing noise from astronomical images using a pixel-specific noise model
C. Burger, B. Schölkopf, S. Harmeling
IEEE Conference on Computational Photography (ICCP), CMU, Pittsburgh, USA, 2011




Computer vision (other than deconvolution and denosing)

Attribute-based classification for zero shot visual object categorization
C. Lampert, H. Nickisch, S. Harmeling
Accepted at IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), 2013

Learning To Detect Unseen Object Classes by Between-Class Attribute Transfer
C. Lampert, H. Nickisch, S. Harmeling
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Miami, USA, 2009

website with datasets



Linear and nonlinear ICA

How to test the quality of reconstructed sources in independent component analysis (ICA) of EEG/MEG Data
M. Grosse-Wentrup, S. Harmeling, T. Zander, J. Hill, B. Schölkopf
Pattern Recognition in Neuro Imaging (PRNI) Workshop, 2013

Inlier-based ICA with an application to superimposed images
F. Meinecke, S. Harmeling, and K.-R. Müller
International Journal of Imaging Systems and Technology (IJIST), Volume 15, Issue 1, pp. 48--55, 2005

Independent Component Analysis and beyond
S. Harmeling
Doktorarbeit (PhD thesis), supervisor Prof. Dr. K.-R. Müller, Universität Potsdam, 2004

Editorial for the special section on ICA
E. Oja, S. Harmeling, and L. Almeida
Signal Processing, Volume 84, pp. 215--216, 2004

Injecting noise for analysing the stability of ICA components
S. Harmeling, F. Meinecke, and K.-R. Müller
Signal Processing, Volume 84, pp. 255--266, 2004

Robust ICA for super-Gaussian sources
F. Meinecke, S. Harmeling, and K.-R. Müller
Fifth International Conference on Independent Component Analysis and Blind Signal Separation (ICA), Granada, Spain, pp. 217--224, 2004

code (ibica-0.1.tar)

Using kernel PCA for initialisation of variational bayesian nonlinear blind source separation method
A. Honkela, S. Harmeling, L. Lundqvist, and H. Valpola
Fifth International Conference on Independent Component Analysis and Blind Signal Separation (ICA), Granada, Spain, pp. 790--797, 2004

Blind separation of post-nonlinear mixtures using linearizing transformations and temporal decorrelation
A. Ziehe, M. Kawanabe, S. Harmeling, and K.-R. Müller
Journal of Machine Learning Research (JMLR), Volume 4 , pp. 1319--1338, 2003

Kernel-based nonlinear blind source separation
S. Harmeling, A. Ziehe, M. Kawanabe, and K.-R. Müller
Neural Computation, Volume 15, pp. 1089--1124, 2003

code (ktdsep-0.2.tar)

Blind separation of post-nonlinear mixtures using gaussianizing transformations and temporal decorrelation
A. Ziehe, M. Kawanabe, S. Harmeling, and K.-R. Müller
Fourth International Symposium on Independent Component Analysis and Blind Signal Separation (ICA), pp. 269--274, Nara, Japan, 2003

Analysing ICA components by injecting noise
S. Harmeling, F. Meinecke, and K.-R. Müller
Fourth International Symposium on Independent Component Analysis and Blind Signal Separation (ICA), pp. 149--154, Nara, Japan, 2003

code (relica-0.1.tar)

Kernel feature spaces and nonlinear blind source separation
S. Harmeling, A. Ziehe, M. Kawanabe, and K.-R. Müller
Conference on Neural Information Processing Systems (NIPS), Volume 14, Vancouver, Canada, 2002

Nonlinear blind source separation using kernel feature spaces
S. Harmeling, A. Ziehe, M. Kawanabe, B. Blankertz, and K.-R. Müller
Third Symposium on Independent Component Analysis and Blind Signal Separation (ICA), pp. 102--107, San Diego, USA, 2001

Separation of post-nonlinear mixtures using ACE and temporal decorrelation
A. Ziehe, M. Kawanabe, S. Harmeling, and K.-R. Müller
Third Symposium on Independent Component Analysis and Blind Signal Separation (ICA), pp. 433--438, San Diego, CA, 2001




Unsupervised learning (other than ICA)

Greedy Learning of Binary Latent Trees
S. Harmeling, C. K. I. Williams
IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), Volume 33, Issue 6, 2011

code (ltt-1.4.tar)

Exploring model selection techniques for nonlinear dimensionality reduction
S. Harmeling
University of Edinburgh, School of Informatics Research Report EDI-INF-RR-0960, 2007

code (nldim-0.1.tar)

From outliers to prototypes: Ordering data
S. Harmeling, G. Dornhege, D. Tax, F. Meinecke, and K.-R. Müller
Neurocomputing, Volume 69, Issues 13-15, pp. 1608--1618, 2006




Supervised learning (see also computer vision)

How to Explain Individual Classification Decisions
D. Baehrens, T. Schroeter, S. Harmeling, M. Kawanabe, K. Hansen, K.-R. Müller
Journal of Machine Learning Research (JMLR), Volume 11, pp. 1803--1931, 2010




Reinforcement learning

Expectation-Maximization methods for solving (PO)MDPs and optimal control problems
M. Toussaint, A. Storkey, S. Harmeling
Book chapter in ``Bayesian Time Series Models'', Edited by D. Barber, A.T. Cemgil, S. Chiappa, Cambridge University Press, 2011

Probabilistic inference for solving MDPs and POMDPs
M. Toussaint, S. Harmeling, A. Storkey
University of Edinburgh, School of Informatics Research Report EDI-INF-RR-0934, 2006




Natural language processing

Inferring Textual Entailment with a Probabistically Sound Calculus
S. Harmeling
Journal of Natural Language Engineering (JNLE), Volume 15, Issue 4, pp. 459--477, 2009

An Extensible Probabilistic Transformation-based Approach to the Third Recognizing Textual Entailment Challenge
S. Harmeling
ACL-PASCAL Workshop on Textual Entailment and Paraphrasing at the Annual Meeting of the Association for Computational Linguistics (ACL), pp. 137-142, Prague, Czech Republic, 2007




Image analysis in cosmology

Image analysis for cosmology: results from the GREAT10 Galaxy Challenge
T.D. Kitching, S.T. Balan, S. Bridle, N. Cantale, F. Courbin, T. Eifler,M. Gentile, M.S.S. Gill, S. Harmeling, C. Heymans, and 18 other authors
Monthly Notices of the Royal Astronomical Society, 423 (4), pp. 3163--3208, 2012

Gravitational Lensing Accuracy Testing 2010 Challenge Handbook
T. Kitching, A. Amara, M. Gill, S. Harmeling, C. Heymans, R. Massey, B. Rowe, T. Schrabback, L. Voigt, S. Balan, G. Bernstein, M. Bethge, S. Bridle, and 22 other authors
Annals of Applied Statistics, 5 (3), 2011

Results of the GREAT08 Challenge: An image analysis competition for cosmological lensing
S. Bridle, S. T. Balan, M. Bethge, M. Gentile, S. Harmeling, C. Heymans, M. Hirsch, R. Hosseini, M. Jarvis, D. Kirk, T. Kitching, K. Kuijken, A. Lewis, S. Paulin-Henriksson, B. Schölkopf, M. Velander, L. Voigt, D. Witherick, and 16 other authors
Monthly Notices of the Royal Astronomical Society, 405 (3), pp. 2044--2061, 2010




Earth system science / Ecoinformatics

Detection and attribution of large spatiotemporal extreme events in Earth observation data
J. Zscheischler, M. Mahecha, S. Harmeling, M. Reichstein
Accepted at Ecological Informatics, 2013

Climate classifications: the value of unsupervised clustering
J. Zscheischler, M.D. Mahecha, S. Harmeling
Third Workshop on Data Mining in Earth System Science (DMESS) 2012




Matrix differential calculus

Matrix Differential Calculus Cheat Sheet
S. Harmeling
Blue Note 142, 2013




Statistics, genetic algorithms

Bayesian Estimators for Robins-Ritov's Problem
S. Harmeling, M. Toussaint
University of Edinburgh, School of Informatics Research Report EDI-INF-RR-1189, 2007

Solving satisfiability problems with genetic algorithms
S. Harmeling
In Genetic Algorithms and Genetic Programming at Stanford 2000, Edited by J.R. Koza, pp. 206--213, 2000
code (gasat-1.1.tar)




Mathematical logic / proof theory

Eine beweistheoretische Anwendung der <k-Relation
S. Harmeling
Diplomarbeit (master thesis), supervisor PD Dr. A. Weiermann, Institut für mathematische Logik und Grundlagenforschung der Universität Münster, 1998




Miscellaneous code

Matlab implementation of kernel PCA
S. Harmeling
code (kpca-0.1.tar)

Matlab implementation of Parra/Spence's blind source separation algorithm
S. Harmeling
code (convbss-0.1.tar)