Header logo is


2012


no image
Expectation-Maximization methods for solving (PO)MDPs and optimal control problems

Toussaint, M., Storkey, A., Harmeling, S.

In Inference and Learning in Dynamic Models, (Editors: Barber, D., Cemgil, A.T. and Chiappa, S.), Cambridge University Press, Cambridge, UK, January 2012 (inbook) In press

ei

PDF [BibTex]

2012


PDF [BibTex]


no image
Inferential structure determination from NMR data

Habeck, M.

In Bayesian methods in structural bioinformatics, pages: 287-312, (Editors: Hamelryck, T., Mardia, K. V. and Ferkinghoff-Borg, J.), Springer, New York, 2012 (inbook)

ei

[BibTex]

[BibTex]


no image
Robot Learning

Sigaud, O., Peters, J.

In Encyclopedia of the sciences of learning, (Editors: Seel, N.M.), Springer, Berlin, Germany, 2012 (inbook)

ei

Web [BibTex]

Web [BibTex]


no image
Reinforcement Learning in Robotics: A Survey

Kober, J., Peters, J.

In Reinforcement Learning, 12, pages: 579-610, (Editors: Wiering, M. and Otterlo, M.), Springer, Berlin, Germany, 2012 (inbook)

Abstract
As most action generation problems of autonomous robots can be phrased in terms of sequential decision problems, robotics offers a tremendously important and interesting application platform for reinforcement learning. Similarly, the real-world challenges of this domain pose a major real-world check for reinforcement learning. Hence, the interplay between both disciplines can be seen as promising as the one between physics and mathematics. Nevertheless, only a fraction of the scientists working on reinforcement learning are sufficiently tied to robotics to oversee most problems encountered in this context. Thus, we will bring the most important challenges faced by robot reinforcement learning to their attention. To achieve this goal, we will attempt to survey most work that has successfully applied reinforcement learning to behavior generation for real robots. We discuss how the presented successful approaches have been made tractable despite the complexity of the domain and will study how representations or the inclusion of prior knowledge can make a significant difference. As a result, a particular focus of our chapter lies on the choice between model-based and model-free as well as between value function-based and policy search methods. As a result, we obtain a fairly complete survey of robot reinforcement learning which should allow a general reinforcement learning researcher to understand this domain.

ei

Web DOI [BibTex]

Web DOI [BibTex]


no image
Higher-Order Tensors in Diffusion MRI

Schultz, T., Fuster, A., Ghosh, A., Deriche, R., Florack, L., Lim, L.

In Visualization and Processing of Tensors and Higher Order Descriptors for Multi-Valued Data, (Editors: Westin, C. F., Vilanova, A. and Burgeth, B.), Springer, 2012 (inbook) Accepted

ei

[BibTex]

[BibTex]

2009


no image
Text Clustering with Mixture of von Mises-Fisher Distributions

Sra, S., Banerjee, A., Ghosh, J., Dhillon, I.

In Text mining: classification, clustering, and applications, pages: 121-161, Chapman & Hall/CRC data mining and knowledge discovery series, (Editors: Srivastava, A. N. and Sahami, M.), CRC Press, Boca Raton, FL, USA, June 2009 (inbook)

ei

Web DOI [BibTex]

2009


Web DOI [BibTex]


no image
Data Mining for Biologists

Tsuda, K.

In Biological Data Mining in Protein Interaction Networks, pages: 14-27, (Editors: Li, X. and Ng, S.-K.), Medical Information Science Reference, Hershey, PA, USA, May 2009 (inbook)

Abstract
In this tutorial chapter, we review basics about frequent pattern mining algorithms, including itemset mining, association rule mining and graph mining. These algorithms can find frequently appearing substructures in discrete data. They can discover structural motifs, for example, from mutation data, protein structures and chemical compounds. As they have been primarily used for business data, biological applications are not so common yet, but their potential impact would be large. Recent advances in computers including multicore machines and ever increasing memory capacity support the application of such methods to larger datasets. We explain technical aspects of the algorithms, but do not go into details. Current biological applications are summarized and possible future directions are given.

ei

Web [BibTex]

Web [BibTex]


no image
Large Margin Methods for Part of Speech Tagging

Altun, Y.

In Automatic Speech and Speaker Recognition: Large Margin and Kernel Methods, pages: 141-160, (Editors: Keshet, J. and Bengio, S.), Wiley, Hoboken, NJ, USA, January 2009 (inbook)

ei

Web [BibTex]

Web [BibTex]


no image
Covariate shift and local learning by distribution matching

Gretton, A., Smola, A., Huang, J., Schmittfull, M., Borgwardt, K., Schölkopf, B.

In Dataset Shift in Machine Learning, pages: 131-160, (Editors: Quiñonero-Candela, J., Sugiyama, M., Schwaighofer, A. and Lawrence, N. D.), MIT Press, Cambridge, MA, USA, 2009 (inbook)

Abstract
Given sets of observations of training and test data, we consider the problem of re-weighting the training data such that its distribution more closely matches that of the test data. We achieve this goal by matching covariate distributions between training and test sets in a high dimensional feature space (specifically, a reproducing kernel Hilbert space). This approach does not require distribution estimation. Instead, the sample weights are obtained by a simple quadratic programming procedure. We provide a uniform convergence bound on the distance between the reweighted training feature mean and the test feature mean, a transductive bound on the expected loss of an algorithm trained on the reweighted data, and a connection to single class SVMs. While our method is designed to deal with the case of simple covariate shift (in the sense of Chapter ??), we have also found benefits for sample selection bias on the labels. Our correction procedure yields its greatest and most consistent advantages when the learning algorithm returns a classifier/regressor that is simpler" than the data might suggest.

ei

PDF Web [BibTex]

PDF Web [BibTex]