Machine Learning

Our research

Our group develops theoretical models and practical technology for the processing of sensory data. We pursue and conduct projects on non-linear dictionary learning, large-scale unsupervised and semi-supervised learning, and inference and learning in neural circuits. Application domains of our learning and inference technology are acoustic data, visual data, and medical data.

We are part of the cluster of excellence Hearing4all and the Department of Medical Physics and Acoustics at the School of Medicine and Health Sciences.

The group was previously located at the TU Berlin and at the Goethe-University Frankfurt am Main. This website is currently under construction. For more details on our research please refer, for now, to our previous website at Frankfurt.

For any inquiries please contact Jörg Lücke.

Selected Recent Publications (Journals)

M. Henniges, R. E. Turner, M. Sahani, J. Eggert, J. Lücke (2014).
Efficient Occlusive Components Analysis.
Journal of Machine Learning Research, in press.

A.-S. Sheikh, J. A. Shelton, J. Lücke (2014).
A Truncated EM Approach for Spike-and-Slab Sparse Coding.
Journal of Machine Learning Research, in press.

Z. Dai and J. Lücke (2014).
Autonomous Document Cleaning – A Generative Approach to Reconstruct Strongly Corrupted Scanned Texts.
IEEE Transactions on Pattern Analysis and Machine Intelligence, in press.

C.-M. Svensson, S. Krusekopf, J. Lücke, M. T. Figge (2014).
Automated Detection of Circulating Tumour Cells With Naive Bayesian Classifiers.
Cytometry Part A 85(6): 501-511.

J. Bornschein, M. Henniges, J. Lücke (2013).
Are V1 simple cells optimized for visual occlusions? A comparative study. (online access, bibtex)
PLOS Computational Biology 9(6): e1003062.

C. Keck*, C. Savin*, J. Lücke (2012).
Feedforward Inhibition and Synaptic Scaling – Two Sides of the Same Coin? (online access, bibtex)
PLOS Computational Biology 8(3): e1002432.
*joint first authorship

J. Lücke and J. Eggert (2010). 
Expectation Truncation and the Benefits of Preselection in Training Generative Models. (pdfbibtexanimations)
Journal of Machine Learning Research 11:2855-2900, 2010.

J. Lücke (2009). 
Receptive Field Self-Organization in a Model of the Fine-Structure in V1 Cortical Columns. (online accessbibtex)
Neural Computation, 21(10):2805-2845.

J. Lücke and M. Sahani (2008). 
Maximal Causes for Non-linear Component Extraction. (pdfbibtex)
Journal of Machine Learning Research 9:1227-1267.

Selected Recent Publications (Conferences)

Z. Dai, G. Exarchakis, and J. Lücke (2013).
What Are the Invariant Occlusive Components of Image Patches? A Probabilistic Generative Approach.
Advances in Neural Information Processing Systems 26, 243-251. (online access)

J. A. Shelton, P. Sterne, J. Bornschein, A.-S. Sheikh,  and J. Lücke (2012).
Why MCA? Nonlinear sparse coding with spike-and-slab prior for neurally plausible image encoding.
Advances in Neural Information Processing Systems 25, 2285-2293. (online access)

Z. Dai and J. Lücke (2012).
Autonomous Cleaning of Corrupted Scanned Documents - A Generative Modeling Approach. (pdf, bibtex)
Proc. IEEE Computer Vision and Pattern Recognition (CVPR), 3338-3345.
(oral presentation and Google Student Travel Award).

Z. Dai and J. Lücke (2012).
Unsupervised Learning of Translation Invariant Occlusive Components. (pdf, bibtex)
Proc. IEEE Computer Vision and Pattern Recognition (CVPR), 2400-2407. 

J. A. Shelton, J. Bornschein, A.-S. Sheikh, P. Berkes, and J. Lücke (2011).
Select and Sample — A Model of Efficient Neural Inference and Learning. (pdfbibtex).
Advances in Neural Information Processing Systems 24, 2618-2626, 2011.

G. Puertas*, J. Bornschein*, and J. Lücke (2010). 
The Maximal Causes of Natural Scenes are Edge Filters. (pdfbibtexsupplement, code)
Advances in Neural Information Processing Systems 23, 1939-1947, 2010.
 *joint first authorship

J. Lücke, R. Turner, M. Sahani, and M. Henniges (2009). 
Occlusive Components Analysis. (pdfbibtexsupplementary)
Advances in Neural Information Processing Systems 22, 1069-1077.

COPYRIGHT NOTICE

The papers listed above have been published after peer review in different journals or conference proceedings. These journals or proceedings remain the only definitive repository of the content. Copyright and all rights therein are usually retained by the respective publishers. These materials may not be copied or reposted without their explicit permission. Use for scholarly purposes only.

 

News

  • 25 April 2014
    We have openings for PhD positions and a postdoc position (see here).

  • 23 April 2014
    The paper "Efficient Occlusive Components Analysis" (Henniges et al.) is accepted by the Journal of Machine Learning Research

  • 15 April 2014
    The paper "A Truncated EM Approach for Spike-and-Slab Sparse Coding" (Sheikh et al.) is accepted by the Journal of Machine Learning Research

  • 26 March 2014
    The paper "Automated Detection of Circulating Tumour Cells With Naive Bayesian Classifiers" (Svensson et al.) is accepted by Cytometry Part A

  • 14 February 2014
    The paper "Autonomous Document Cleaning – A Generative Approach to Reconstruct Strongly Corrupted Scanned Texts" (Dai & Lücke) is accepted by IEEE Transactions on Pattern Analysis and Machine Intelligence

  • 17 December 2013
    Jörg Bornschein (now CIFAR fellow at the University of Montreal) successfully defended his PhD Thesis at Frankfurt - Congratulations!

  • 12 November 2013
    Zhenwen Dai (now postdoc at the lab of Neil Lawrence, Sheffield, UK) successfully defended his PhD Thesis at Frankfurt - Congratulations!

  • 12 November 2013
    Christian Keck (now software engineer and consultant) successfully defended his PhD Thesis at Bochum - Congratulations!

  • 5 September 2013
    The paper "What Are the Invariant Occlusive Components of Image Patches? A Probabilistic Generative Approach" (Dai et al.) is accepted by the NIPS conference

  • 21 March 2013
    The paper "Are V1 simple cells optimized for visual occlusions? A comparative study" (Bornschein et al.) is accepted by PLOS Computational Biology

  • 10 January 2013
    Marc Henniges (now consultant for risk management) successfully defended his PhD Thesis at Frankfurt - Congratulations!