A Spectral Analysis of Dot-product Kernels

被引:0
|
作者
Scetbon, Meyer [1 ]
Harchaoui, Zaid [2 ]
机构
[1] CREST, ENSAE, Palaiseau, France
[2] Univ Washington, Seattle, WA 98195 USA
关键词
INTEGRAL-OPERATORS; REGULARIZATION; CLASSIFICATION;
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We present eigenvalue decay estimates of integral operators associated with compositional dot-product kernels. The estimates improve on previous ones established for power series kernels on spheres. This allows us to obtain the volumes of balls in the corresponding reproducing kernel Hilbert spaces. We discuss the consequences on statistical estimation with compositional dot product kernels and highlight interesting trade-offs between the approximation error and the statistical error depending on the number of compositions and the smoothness of the kernels.
引用
收藏
页数:11
相关论文
共 50 条
  • [31] DOT-PRODUCT BASED GLOBAL AND LOCAL FEATURE FUSION FOR IMAGE SEARCH
    Hu, Zechao
    Bors, Adrian G.
    [J]. 2022 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING, ICIP, 2022, : 1911 - 1915
  • [32] A Scalable BFloat16 Dot-Product Architecture for Deep Learning
    Zhang, Jing
    Huang, Libo
    Tan, Hongbing
    Zheng, Zhong
    Guo, Hui
    [J]. PROCEEDINGS OF THE GREAT LAKES SYMPOSIUM ON VLSI 2023, GLSVLSI 2023, 2023, : 219 - 220
  • [33] Optical coherent dot-product chip for sophisticated deep learning regression
    Xu, Shaofu
    Wang, Jing
    Shu, Haowen
    Zhang, Zhike
    Yi, Sicheng
    Bai, Bowen
    Wang, Xingjun
    Liu, Jianguo
    Zou, Weiwen
    [J]. LIGHT-SCIENCE & APPLICATIONS, 2021, 10 (01)
  • [34] Conditionally positive definite dot product kernels
    Menegatto, V. A.
    Oliveira, C. P.
    Peron, A. P.
    [J]. JOURNAL OF MATHEMATICAL ANALYSIS AND APPLICATIONS, 2006, 321 (01) : 223 - 241
  • [35] On conditionally positive definite dot product kernels
    Menegatto, V. A.
    Oliveira, C. P.
    Peron, Ana P.
    [J]. ACTA MATHEMATICA SINICA-ENGLISH SERIES, 2008, 24 (07) : 1127 - 1138
  • [36] On Conditionally Positive Definite Dot Product Kernels
    V.A.MENEGATTO
    C.P.OLIVEIRA
    Ana P.PERON
    [J]. Acta Mathematica Sinica,English Series, 2008, 24 (07) : 1127 - 1138
  • [37] Improved Random Features for Dot Product Kernels
    Wacker, Jonas
    Kanagawa, Motonobu
    Filippone, Maurizio
    [J]. JOURNAL OF MACHINE LEARNING RESEARCH, 2024, 25 : 1 - 75
  • [38] On conditionally positive definite dot product kernels
    V. A. Menegatto
    C. P. Oliveira
    Ana P. Peron
    [J]. Acta Mathematica Sinica, English Series, 2008, 24 : 1127 - 1138
  • [39] SFDoP: A Scalable Fused BFloat16 Dot-Product Architecture for DNN
    Zhang, Jing
    Tan, Hongbing
    Huang, Libo
    [J]. 2023 IEEE 41ST INTERNATIONAL CONFERENCE ON COMPUTER DESIGN, ICCD, 2023, : 62 - 65
  • [40] Radius-Margin Ratio Optimization for Dot-Product Boolean Kernel Learning
    Lauriola, Ivano
    Polato, Mirko
    Aiolli, Fabio
    [J]. ARTIFICIAL NEURAL NETWORKS AND MACHINE LEARNING, PT II, 2017, 10614 : 183 - 191