Sciweavers

1157 search results - page 160 / 232
» A kernel path algorithm for support vector machines
Sort
View
AIME
2009
Springer
14 years 29 days ago
The Role of Biomedical Dataset in Classification
In this paper, we investigate the role of a biomedical dataset on the classification accuracy of an algorithm. We quantify the complexity of a biomedical dataset using five complex...
Ajay Kumar Tanwani, Muddassar Farooq
NIPS
2007
13 years 10 months ago
A Risk Minimization Principle for a Class of Parzen Estimators
This paper1 explores the use of a Maximal Average Margin (MAM) optimality principle for the design of learning algorithms. It is shown that the application of this risk minimizati...
Kristiaan Pelckmans, Johan A. K. Suykens, Bart De ...
ICML
2005
IEEE
14 years 9 months ago
Supervised versus multiple instance learning: an empirical comparison
We empirically study the relationship between supervised and multiple instance (MI) learning. Algorithms to learn various concepts have been adapted to the MI representation. Howe...
Soumya Ray, Mark Craven
IBPRIA
2009
Springer
14 years 1 months ago
Class Representative Visual Words for Category-Level Object Recognition
Recent works in object recognition often use visual words, i.e. vector quantized local descriptors extracted from the images. In this paper we present a novel method to build such ...
Roberto Javier López-Sastre, Tinne Tuytelaa...
ICIP
2003
IEEE
14 years 10 months ago
Classification-based spatial error concealment for images
* This paper presents a new, classification-based spatial error concealment algorithm for images. The proposed scheme takes advantage of two state-of-the-art concealment schemes an...
M. Chen, M. Wu, Y. Zheng