About

This is the research home page of Dr. Frank-Michael Schleif. I am currently a PostDoc Researcher at the University Leipzig, AG Computational Intelligence and a Research Associate of Bruker Bioscience Corp. I am interested in algorithms for high dimensional data sets, with a special but not exclusive focus on biomedical data, especially in the question how to efficiently generate interpretable classification models and data visualizations. Thereby prototype based approaches are most important in my research.

In my Ph.D. thesis I developed (among other things) a new method to efficiently generate prototype based classification models for data sets of clinical proteom MS spectra. Now, I develop further generalizations of these approaches under the light of different domain specific metrics, new methods for confidence and outlier estimations, visualization and data preprocessing as well as efficient data processing by means of appropriate data structures.

Personal

		since 11/2006	Postdoctoral Research Fellow and Lecturer with PD Dr. rer. nat Thomas Villmann Medical Department Leipzig University Since 01/2007 I am part project leader in the project Biodiversity Since 03/2009 I am part project leader in the project MetaSTEM
		2006	Ph.D. in Computer Science (Dr. rer. nat.) at Clausthal University of Technology Thesis entitled Prototype based Machine Learning in Clinical Proteomics
		2004-2006	Researcher and Software Developer at Bruker Daltonics (Numerical methods team) and group of bio-analytics (external) PhD student at CIG Clausthal University of Technology (Prof. Barbara Hammer
		2003	Research Fellow with Prof Dr. rer. nat Volker Gruhn Computer Science Department, Leipzig University
		2002	Diploma in Computer Science at Leipzig University Thesis entitled Moment based methods for optical character recognition

Research interests

Prototype based data analysis

Learning Vector Quantizers (LVQ)
Supervised Relevance LVQ and improvements
Fuzzy approaches for supervised and unsupervised vector quantization
Visualization and analysis of high dimensional dataspaces


Enhancement and theoretical analysis of prototype networks Metric adaptation in LVQ in very high dimensional dataspaces Algorithms for feature extraction of LCMS measurements, MS, ..., OCR Determination of relevant input dimensions / feature selection (SRNG, wrapper methods, Genetic algorithms) Rule extraction Cost functions for margin optimizers Generalisation theory



Applications and method development for bioinformatics, analysis of spectral data (MALDI-MS,IMS,LC-MS,NMR,satelite remote sensing). Development of preprocessing and high level analysis algorithms, process optimization, statistical modeling, theoretical analysis for signals and images. Clinical proteomics (MS,LCMS,Tissue-MS) Metabolomics (H-NMR,13C-NMR) Chemometrics (IMS - hazardous material detection) Tissue, slice analysis and modeling Bacterial analysis (identification) Signal and Image processing biomarker discovery

Cooperations

Research can not be done alone - so I am lucky to have some fellows working together with me on some projects.

Bruker Daltonik GmbH - much support for my PhD thesis project mainly related to improvements in Learning Vector Quantization
Rice University - in coop. with Dr. Th. Villmann, I am involved in some analysis of high dimensional data and algorithm development
IPK Gatersleben - cooperation in development of algorithms for fuzzy classification of fuzzy labeled data
University of Clausthal - coop. in different topics of (un)supervised methods for data analysis of proteomic data
University of Bamberg - coop. in data analysis with LVQ methods for psychological studies
Computational Intelligence Group @ University Leipzig - I am currently working on a project of NMR spectroscopy in the analysis of stem cells

Further research contacts

Tal Ronen, Viral Dynamics Modeling Laboratory, Bar Ilan University, Israel

Session/Conference organization/Reviewer activities

We organize a special session at the ESANN 2010
For details please take a look at the call for papers: ESANN Special Session on sparse representation of data

I am co-organizer of the 1st Mittweidaer Workshop on Computational Intelligence MIWOCI 2009

We organized a special session at the ESANN 2009 - ESANN Special Session on Neural Maps and Learning Vector Qunatization - Theory and Applications Thanks to all who submitted a contribution.

We organized a special session at the CBMS 2008 Thanks to all who submitted a contribution.
For details please take a look at the call for papers: CBMS Special Session on Machine Learning Methods for High-Dimensional Data in Bio-medicine

We organized a special session at the FLINS 2006. Thanks to all who submitted a contribution.
For details please take a look at the call for papers: FLINS Special Session on Data Analysis for Mass Spectrometric Problems

Conferences

ICSE 2004
ESANN 2005, 2006, 2007, 2008, 2009 (Special session)
FLINS 2006 (Special session)
ICMLA 2007
CBMS 2008 (Special session), 2009
CIBB 2008
MICAI 2008, 2009
ICANN 2009

Journal

Pattern Recognition
Pattern Recognition Letters
Pattern Analysis and Applications
Neurocomputing
Neurocomputing - Special Issue ESANN 2008, ESANN 2009 (Guest editor)
Neurocomputing Letters
Neural Networks
IEEE-TNN
Information Sciences
Artificial Intelligence in Medicine
Advances in Fuzzy Systems
Knowledge and Information Systems

PC member

CBMS 2008, 2009
CIBB 2008
AIA 2010
SNPD 2010

Reviews other

(book) Applications of Computational Intelligence in Bioinformatics and Biomedicine: Current Trends and Open Problems

Awards/Nominations:

Poster award - Research Festival 2004 - University of Leipzig, Germany

Best Dissertation in Computer Science - Technical University of Clausthal 2006

Nominated for GI Dissertationspreis 2006

Projects/Fundings

[2004--current] pattern recognition in clinical proteomics (industrial funding [bruker bioscience], german research foundation)
Results: multiple research papers in journals and conferences, one patent, data analysis package ClinProTools 2.0 - 2.2
1 research associate, travel grants (~120.000 EUR)

[11/2006--current] pattern recognition and data analysis for NMR in analysis of stem cell data (BMBF funding)
Results: initial results in theoretic extensions of prototype data analysis methods
1 research associate, student assistant fundings (~150.000 EUR)

[2008--current] project leader - project: biodiversity - research and improvements for identification of bacterial data by mass spectrometry (SAB and bruker bioscience funding)
Results: project started in Januar 2008, initial results on outlier detection and (semi)-supervised model confidence estimations
1 PhD position, travel grants (~50.000 EUR)

[2009--current] part project leader - project: single cell imaging - research and improvements for prototype based classifiers to obtain models to represent very large set of MALDI-imaging spectra on single cell resolution. Improvements to model uncertainty of class assignments (Industrial funding Bruker Daltonik GmbH, managed by University of Applied Sciences, Mittweida)
1 PhD position, travel grants (~40.000 EUR)

Next events:

You can find me at ESANN 2010, Bruges, B, Statistics and the Life Science, Groningen, NL