Sciweavers

1854 search results - page 4 / 371
» On Kernel Methods for Relational Learning
Sort
View
TCS
2008
13 years 7 months ago
Kernel methods for learning languages
This paper studies a novel paradigm for learning formal languages from positive and negative examples which consists of mapping strings to an appropriate highdimensional feature s...
Leonid Kontorovich, Corinna Cortes, Mehryar Mohri
ICML
2006
IEEE
14 years 8 months ago
Practical solutions to the problem of diagonal dominance in kernel document clustering
In supervised kernel methods, it has been observed that the performance of the SVM classifier is poor in cases where the diagonal entries of the Gram matrix are large relative to ...
Derek Greene, Padraig Cunningham
JMLR
2010
147views more  JMLR 2010»
13 years 2 months ago
Image Denoising with Kernels Based on Natural Image Relations
A successful class of image denoising methods is based on Bayesian approaches working in wavelet representations. The performance of these methods improves when relations among th...
Valero Laparra, Juan Gutierrez, Gustavo Camps-Vall...
CVPR
2007
IEEE
14 years 9 months ago
Connecting the Out-of-Sample and Pre-Image Problems in Kernel Methods
Kernel methods have been widely studied in the field of pattern recognition. These methods implicitly map, "the kernel trick," the data into a space which is more approp...
Pablo Arias, Gregory Randall, Guillermo Sapiro
AAAI
2007
13 years 10 months ago
Recognizing Textual Entailment Using a Subsequence Kernel Method
We present a novel approach to recognizing Textual nt. Structural features are constructed from abstract tree descriptions, which are automatically extracted from syntactic depend...
Rui Wang 0005, Günter Neumann