A Unified View of Localized Kernel Learning.
John Moeller,Sarathkrishna Swaminathan,Suresh Venkatasubramanian +2 more
- pp 252-260
Reads0
Chats0
TLDR
Localized Kernel Learning (LKL) as discussed by the authors is an extension of multiple kernel learning (MKL) which aims to learn not only a classifier/regressor but also the best kernel for the training task, usually from a combination of existing kernel functions.Abstract:
Multiple Kernel Learning, or MKL, extends (kernelized) SVM by attempting to learn not only a classifier/regressor but also the best kernel for the training task, usually from a combination of existing kernel functions. Most MKL methods seek the combined kernel that performs best over every training example, sacrificing performance in some areas to seek a global optimum. Localized kernel learning (LKL) overcomes this limitation by allowing the training algorithm to match a component kernel to the examples that can exploit it best. Several approaches to the localized kernel learning problem have been explored in the last several years. We unify many of these approaches under one simple system and design a new algorithm with improved performance. We also develop enhanced versions of existing algorithms, with an eye on scalability and performance.read more
Citations
More filters
Journal ArticleDOI
Optimizing Kernel Machines Using Deep Learning
TL;DR: The idea of using deep architectures to perform kernel machine optimization, for both computational efficiency and end-to-end inferencing, is explored and the kernel dropout regularization is introduced to enable improved training convergence.
Journal ArticleDOI
A three-level Multiple-Kernel Learning approach for soil spectral analysis
Nikolaos L. Tsakiridis,Christos G. Chadoulos,John B. Theocharis,Eyal Ben-Dor,George C. Zalidis +4 more
TL;DR: The proposed MKL framework was compared with other state-of-the-art approaches, and the results indicated that it attains the best performance in terms of accuracy, whilst at the same time producing interpretable results.
Journal ArticleDOI
Soft-clustering-based local multiple kernel learning algorithm for classification
TL;DR: A soft-clustering-based local multiple kernel learning algorithm to tackle the issues above and indicates the kernel weights solved by the algorithm are better suitable for the characteristics of the dataset.
Journal ArticleDOI
Localized multiple kernel learning using graph modularity
TL;DR: In this article , a graph-based heuristic approach for multiple kernel learning (MKL) is proposed, which assigns sample-specific kernel weights based on contribution to graph modularity.