Titlebook: Covariances in Computer Vision and Machine Learning; Hà Quang Minh,Vittorio Murino Book 2018 Springer Nature Switzerland AG 2018

Peculate

stances and divergences between them, we now discuss some of the most important problems encountered in practical applications, namely classification and regression on SPD matrices. In machine learning, a prominent paradigm for solving classification and regression problems is that of kernel methods

aesthetician

is chapter, by employing the feature map viewpoint of kernel methods in machine learning, we generalize covariance matrices to infinite-dimensional covariance operators in RKHS. Since they encode . between input features, they can be employed as a powerful form of data representation, which we explo

agenda

Merited

an distance, and Log-Hilbert-Schmidt distance and inner product between RKHS covariance operators. In this chapter, we show how the Hilbert-Schmidt and Log-Hilbert-Schmidt distances and inner products can be used to define positive definite kernels, allowing us to apply kernel methods on top of cova

Noisome

exclusice

978-3-031-00692-0Springer Nature Switzerland AG 2018

幾何學(xué)家

不近人情

an distances and divergences intrinsic to SPD matrices, as described in Chapter 2, it is necessary to define new positive definite kernels based on these distances and divergences. In this chapter, we describe these kernels and the corresponding kernel methods.

調(diào)情

model . in the input data, can substantially outperform finite-dimensional covariance matrices, which only model . in the input. This performance gain comes at higher computational costs and we showed how to substantially decrease these costs via approximation methods.

FAST

Kernel Methods on Covariance Matricesan distances and divergences intrinsic to SPD matrices, as described in Chapter 2, it is necessary to define new positive definite kernels based on these distances and divergences. In this chapter, we describe these kernels and the corresponding kernel methods.