Open AccessProceedings Article
Kernel matrix evaluation
Canh Hao Nguyen,Tu Bao Ho +1 more
- 06 Jan 2007
- pp 987-992
TL;DR: This work proposes an efficient surrogate measure to evaluate the goodness of a kernel matrix based on the data distributions of classes in the feature space that not only overcomes the limitations of KTA, but also possesses other properties like invariance, efficiency and error bound guarantee.
read more
Abstract: We study the problem of evaluating the goodness of a kernel matrix for a classification task. As kernel matrix evaluation is usually used in other expensive procedures like feature and model selections, the goodness measure must be calculated efficiently. Most previous approaches are not efficient, except for Kernel Target Alignment (KTA) that can be calculated in O(n2) time complexity. Although KTA is widely used, we show that it has some serious drawbacks. We propose an efficient surrogate measure to evaluate the goodness of a kernel matrix based on the data distributions of classes in the feature space. The measure not only overcomes the limitations of KTA, but also possesses other properties like invariance, efficiency and error bound guarantee. Comparative experiments show that the measure is a good indication of the goodness of a kernel matrix.
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
Citations
Simple but effective methods for combining kernels in computational biology
H. Tanabe,Tu Bao Ho,Canh Hao Nguyen,Saori Kawasaki +3 more
- 13 Jul 2008
TL;DR: Two simple but effective methods for determining weights for conic combination of multiple kernels are proposed, to learn optimal weights formulated by the measure FSM for kernel matrix evaluation (feature space-basedkernel matrix evaluation measure), denoted by FSM-MKL.
46
Optimizing the Gaussian kernel function with the formulated kernel target alignment criterion for two-class pattern classification
TL;DR: A novel fast method to optimize the Gaussian kernel function for two-class pattern classification tasks, where it is desirable for the kernel machines to use an optimized kernel that adapts well to the input data and the learning tasks.
45
Learning by local kernel polarization
TL;DR: A new quality measure called local kernel polarization is proposed, which is a localized variant of kernel polarization that can preserve the local structure of the data of the same class so the data can be embedded more appropriately.
35
Patent
Classifying objects in a scene
Xiaoxu Ma,Lingyun Liu,Daniel Joseph Filip,Luc Vincent,Christopher Richard Uhlik +4 more
- 25 May 2007
TL;DR: In this article, a computer-implemented method of classifying image data includes receiving a plurality of data points corresponding to three-dimensional image data, creating from the plurality of points a first subset of points that are above a ground plane in a scene represented by the plurality, identifying a second subset of objects associated with an object in the scene from the first subset, and determining a signature for the identified plurality of features.
33
•Proceedings Article
Eigenvalues ratio for kernel selection of kernel methods
Yong Liu,Shizhong Liao +1 more
- 25 Jan 2015
TL;DR: A novel measure, called eigenvalues ratio (ER), of the tight bound of generalization error for kernel selection is proposed, which is the ratio between the sum of the main eigen Values and that of the tail eigen values of the kernel matrix.
24
References
•Book
Table of Integrals, Series, and Products
I.S. Gradshteyn,I.M. Ryzhik,Alan Jeffrey,Y. V. Geronimus,M. Y. Tseytlin,Y. C. Fung +5 more
- 01 Jan 1943
TL;DR: Combinations involving trigonometric and hyperbolic functions and power 5 Indefinite Integrals of Special Functions 6 Definite Integral Integral Functions 7.Associated Legendre Functions 8 Special Functions 9 Hypergeometric Functions 10 Vector Field Theory 11 Algebraic Inequalities 12 Integral Inequality 13 Matrices and related results 14 Determinants 15 Norms 16 Ordinary differential equations 17 Fourier, Laplace, and Mellin Transforms 18 The z-transform
28.8K
An introduction to probability theory and its applications - 3/E. volume 3
William Feller
- 22 Mar 2002
Abstract: The classic text for understanding complex statistical probability An Introduction to Probability Theory and Its Applications offers comprehensive explanations to complex statistical problems. Delving deep into densities and distributions while relating critical formulas, processes and approaches, this rigorous text provides a solid grounding in probability with practice problems throughout. Heavy on application without sacrificing theory, the discussion takes the time to explain difficult topics and how to use them. This new second edition includes new material related to the substitution of probabilistic arguments for combinatorial artifices as well as new sections on branching processes, Markov chains, and the DeMoivreLaplace theorem.
21.5K
A table of integrals
Hudson, Ralph G. , b.
- 01 Jan 1917
TL;DR: Basic Forms x n dx = 1 n + 1 x n+1 (1) 1 x dx = ln |x| (2) udv = uv − vdu (3) 1 ax + bdx = 1 a ln|ax + b| (4) Integrals of Rational Functions
12K
Related Papers (5)
Nello Cristianini,John Shawe-Taylor,André Elisseeff,Jaz S. Kandola +3 more
- 03 Jan 2001
John Shawe-Taylor,Nello Cristianini +1 more
- 01 Jan 2004
Vladimir Vapnik
- 01 Jan 1995