Scaled rotation regularization
نویسنده
چکیده
A new regularization method-a scaled rotation-is proposed and compared with the standard linear regularized discriminant analysis. A sense of the method consists in the singular value decomposition S؍TDT of a sample covariance matrix S and a use of the following representation of an inverse of the covariance matrix S\"T?(D#I)\T?. For certain data structures the scaled rotation helps to reduce the generalization error in small learning-set and high dimensionality cases. E$cacy of the scaled rotation increases if one transforms the data by y"(D#I)\T?x and uses an optimally stopped single layer perceptron classi"er afterwards. 2000 Pattern Recognition Society. Published by Elsevier Science Ltd. All rights reserved.
منابع مشابه
Adaptive Total Variation Based on Feature Scale
The widely used Total Variation de-noising algorithm can preserve sharp edge, while removing noise. However, since fixed regularization parameter over entire image, small details and textures are often lost in the process. In this paper, we propose a modified Total Variation algorithm to better preserve smaller-scaled features. This is done by allowing an adaptive regularization parameter to co...
متن کاملOptimally scaled vector regularization method to solve ill-posed linear problems
Instead of the Tikhonov regularization method which with a scalar being the regularization parameter, Liu et al. [1] have proposed a novel regularization method with a vector as being the regularization parameter. As a continuation we further propose an optimally scaled vector regularization method (OSVRM) to solve the ill-posed linear problems, which is better than the Tikhonov regularization ...
متن کاملEdge-preserving and scale-dependent properties of total variation regularization
Abstract We give and prove two new and fundamental properties of total-variationminimizing function regularization (TV regularization): edge locations of function features tend to be preserved, and under certain conditions are preserved exactly; intensity change experienced by individual features is inversely proportional to the scale of each feature. We give and prove exact analytic solutions ...
متن کاملEvaluation of New MOND Interpolating Function with Rotation Curves of Galaxies
The rotation curves of a sample of 46 low- and high-surface brightness galaxies are considered in the context of Milgrom's modi_ed dynamics (MOND) to test a new interpolating function proposed by Zhao et al. (2010) [1] and compare with the results of simple interpolating function. The predicted rotation curves are calculated from the total baryonic matter based on the B-band surface photometry,...
متن کاملSupervised Scaled Regression Clustering: An Alternative to Neural Networks
This paper describes a rather novel method for the supervised training of regression systems that can be an alternative to feedforward Artificial Neural Networks (ANNs) trained with the BackPropagation algorithm. The proposed methodology is a hybrid structure based on supervised clustering with genetic algorithms and local learning. Supervised Scaled Regression Clustering with Genetic Algorithm...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
- Pattern Recognition
دوره 33 شماره
صفحات -
تاریخ انتشار 2000