About
This workshop aims at collecting theoretical insights in the design of data-dependent learning strategies. Specifically we are interested in how far learned prediction rules may be characterized in terms of the observations themselves. This amounts to capturing how well data can be used to construct structured hypothesis spaces for risk minimization strategies - termed empirical hypothesis spaces. Classical analysis of learning algorithms requires the user to define a proper hypothesis space before seeing the data. In practice however, one often decides on the proper learning strategy or the form of the prediction rules of interest after inspection of the data. This theoretical gap constitutes exactly the scope of this workshop.
More information about the workshop can be found here.
Videos

Online Prediction on Large Diameter Graphs
Dec 20, 2008
·
3318 views

From On-line Algorithms to Data-Dependent Generalization
Dec 20, 2008
·
3008 views

Chromatic PAC-Bayes Bounds for Non-IID Data
Dec 20, 2008
·
3131 views

Representation of Prior Knowledge - from Bias to 'Meta-Bias'
Dec 20, 2008
·
3087 views

Transductive Learning and Computer Vision
Dec 20, 2008
·
4116 views

Exploiting Cluster Structure to Predict The Labeling of a Graph
Dec 20, 2008
·
3006 views

Generalization Bounds for Indefinite Kernel Machines
Dec 20, 2008
·
4114 views

Sample Complexity for Multiresolution ICA
Dec 20, 2008
·
2853 views

Semi-Supervised Learning and Learning via Similarity Functions: two key settings...
Dec 20, 2008
·
8484 views

Online Graph Prediction with Random Trees
Dec 20, 2008
·
2335 views

The use of Unlabeled Data in Supervised Learning: the Manifold Dossier
Dec 20, 2008
·
3198 views

Study of Classification Algorithms using Moment Analysis
Dec 20, 2008
·
3313 views

Theory of Matching Pursuit in Kernel Defined Feature Spaces
Dec 20, 2008
·
4872 views