Validation in Statistics and Machine Learning

6-7 October 2010


Wednesday, October 6
09.00-09.15Welcome (slides)
Session: Telling the Story (Chair: Nicole Krämer)
09.15-10.00Anne Laure Boulesteix (Munich)
Over-optimism in statistical bioinformatics: an illustration (abstract, slides)
10.00-10.45Neil Lawrence (Sheffield)
Making Implementations Available for the Research Community (abstract, slides)
10.45-11.15Coffee Break
Session: Variable Importance (Chair: Carolin Strobl)
11.15-12.00Ulrike Grömping (Berlin)
Variable Importance in Linear Models and Random Forests (abstract)
12.00-12.20Verena Zuber (Leipzig)
High-dimensional feature selection by decorrelation (abstract, slides)
12.20-14.00Lunch Break
Session: Reproducibility (Chair: Neil Lawrence)
14.00-14.45Niels Keiding (Copenhagen)
Reproducible research and the substantive context (abstract, slides)
14.45-15.30Torsten Hothorn (Munich)
Reproducible Statistical Analyses Today (abstract, slides)
15.30-17.00Coffee Break and Poster Session
Poster Session (room 406)
Jean-Charles Lamirel (Vandouevre-les-Nancy) and Ghada Safi (Aleppo)
Use of distance-based indexes might well lead to misinterpretation of clustering quality results (abstract)
Andreas Mayr (Erlangen)
The correct validation of prediction intervals (abstract)
Renee Menezes (Amsterdam)
Filtering, FDR and bias in high-dimensional data analysis (abstract)
Hans-Joachim Mucha (Berlin)
Validation in cluster analysis (abstract)
Tsuyoshi Okita (Dublin)
Statistical Significance Test in Machine Translation (abstract)
Julia Schiffner (Dortmund)
Bias-Variance Analysis of Local Classification Methods (abstract)
Francois Schnitzler (Liege)
Discussing the validation of high-dimensional probability distribution learning with mixtures of graphical models for inference (abstract)
Caroline Truntzer (Dijon)
Comparative optimism in models involving both classical clinical and gene expression information (abstract)
Session: Assessing Prediction Models (Chair: Torsten Hothorn)
17.00-17.45Jelle Goeman (Leiden)
Fast approximate leave-one-out cross-validation for large sample sizes (abstract, slides)
17.45-18.30Thomas A. Gerds (Copenhagen)
Confidence scores for prediction models (abstract, slides)
20.00-24.00Conference Dinner at Tapas y Mas Tres(link)
Thursday, October 7
Session: Variable Importance and Tests (Chair: Ulrike Grömping)
09.00-09.45Carolin Strobl (Munich)
What we can learn from trees and forests (abstract, slides)
09.45-10.30Thorsten Dickhaus (Berlin)
How normal can the t-statistic possibly be? (abstract, slides)
10.30-11.00Coffee Break
Session: Biometry (Chair: Jelle Goeman)
11.00-11.45Ulrich Mansmann (Munich)
Biological aspects for the validation of estimated gene interaction networks from microarray data (abstract, slides)
11.45-12.05Christoph Bernau (Munich)
Correction for Tuning Bias in Resampling Based Error Rate Estimation (abstract, slides)
12.05-12.50Richardus Vonk (Berlin)
The Many Faces of Validation (abstract, slides)
12.50-14.30Lunch Break
Session: Benchmarking (Chair: Paul Eilers)
14.30-14.50Manuel Eugster (Munich)
Sequential Benchmarking (abstract, slides)
14.50-15.35Francois Fleuret (Martigny)
The MASH platform - Collaborative design of large-scale learning architectures (abstract, slides)
15.35-16.20Mikio Braun (Berlin)
Machine Learning Open Source Software and Benchmark Repository (abstract, slides)
16.20-16.30Farewell (slides)