The Experts below are selected from a list of 39726 Experts worldwide ranked by ideXlab platform
Kenji Fukumizu - One of the best experts on this subject based on the ideXlab platform.
-
Model-based kernel Sum Rule: kernel Bayesian inference with probabilistic models
Machine Learning, 2020Co-Authors: Yu Nishiyama, Motonobu Kanagawa, Arthur Gretton, Kenji FukumizuAbstract:Kernel Bayesian inference is a principled approach to nonparametric inference in probabilistic graphical models, where probabilistic relationships between variables are learned from data in a nonparametric manner. Various algorithms of kernel Bayesian inference have been developed by combining kernelized basic probabilistic operations such as the kernel Sum Rule and kernel Bayes’ Rule. However, the current framework is fully nonparametric, and it does not allow a user to flexibly combine nonparametric and model-based inferences. This is inefficient when there are good probabilistic models (or simulation models) available for some parts of a graphical model; this is in particular true in scientific fields where “models” are the central topic of study. Our contribution in this paper is to introduce a novel approach, termed the model-based kernel Sum Rule (Mb-KSR), to combine a probabilistic model and kernel Bayesian inference. By combining the Mb-KSR with the existing kernelized probabilistic Rules, one can develop various algorithms for hybrid (i.e., nonparametric and model-based) inferences. As an illustrative example, we consider Bayesian filtering in a state space model, where typically there exists an accurate probabilistic model for the state transition process. We propose a novel filtering method that combines model-based inference for the state transition process and data-driven, nonparametric inference for the observation generating process. We empirically validate our approach with synthetic and real-data experiments, the latter being the problem of vision-based mobile robot localization in robotics, which illustrates the effectiveness of the proposed hybrid approach.
-
model based kernel Sum Rule kernel bayesian inference with probabilistic models
arXiv: Machine Learning, 2014Co-Authors: Yu Nishiyama, Motonobu Kanagawa, Arthur Gretton, Kenji FukumizuAbstract:Kernel Bayesian inference is a principled approach to nonparametric inference in probabilistic graphical models, where probabilistic relationships between variables are learned from data in a nonparametric manner. Various algorithms of kernel Bayesian inference have been developed by combining kernelized basic probabilistic operations such as the kernel Sum Rule and kernel Bayes' Rule. However, the current framework is fully nonparametric, and it does not allow a user to flexibly combine nonparametric and model-based inferences. This is inefficient when there are good probabilistic models (or simulation models) available for some parts of a graphical model; this is in particular true in scientific fields where "models" are the central topic of study. Our contribution in this paper is to introduce a novel approach, termed the {\em model-based kernel Sum Rule} (Mb-KSR), to combine a probabilistic model and kernel Bayesian inference. By combining the Mb-KSR with the existing kernelized probabilistic Rules, one can develop various algorithms for hybrid (i.e., nonparametric and model-based) inferences. As an illustrative example, we consider Bayesian filtering in a state space model, where typically there exists an accurate probabilistic model for the state transition process. We propose a novel filtering method that combines model-based inference for the state transition process and data-driven, nonparametric inference for the observation generating process. We empirically validate our approach with synthetic and real-data experiments, the latter being the problem of vision-based mobile robot localization in robotics, which illustrates the effectiveness of the proposed hybrid approach.
Kweichou Yang - One of the best experts on this subject based on the ideXlab platform.
-
form factors ofbu d sdecays intop wave axial vector mesons in the light cone Sum Rule approach
Physical Review D, 2008Co-Authors: Kweichou YangAbstract:We calculate the vector and axial-vector form factors of B{sub u,d,s} decays into P-wave axial-vector mesons in the light-cone Sum Rule approach. For the Sum Rule results, we have included corrections of order m{sub A}/m{sub b}, where m{sub A} is the mass of the axial-vector meson A. The results are relevant to the light-cone distribution amplitudes of the axial-vector mesons. It is important to note that, owing to the G parity, the chiral-even two-parton light-cone distribution amplitudes of the {sup 3}P{sub 1} ({sup 1}P{sub 1}) mesons are symmetric (antisymmetric) under the exchange of quark and antiquark momentum fractions in the SU(3) limit. For chiral-odd light-cone distribution amplitudes, it is the other way around. The predictions for decay rates of B{sub u,d,s}{yields}Ae{nu}{sub e} are also presented.
-
form factors of b u d s decays into p wave axial vector mesons in the light cone Sum Rule approach
Physical Review D, 2008Co-Authors: Kweichou YangAbstract:We calculate the vector and axial-vector form factors of B{sub u,d,s} decays into P-wave axial-vector mesons in the light-cone Sum Rule approach. For the Sum Rule results, we have included corrections of order m{sub A}/m{sub b}, where m{sub A} is the mass of the axial-vector meson A. The results are relevant to the light-cone distribution amplitudes of the axial-vector mesons. It is important to note that, owing to the G parity, the chiral-even two-parton light-cone distribution amplitudes of the {sup 3}P{sub 1} ({sup 1}P{sub 1}) mesons are symmetric (antisymmetric) under the exchange of quark and antiquark momentum fractions in the SU(3) limit. For chiral-odd light-cone distribution amplitudes, it is the other way around. The predictions for decay rates of B{sub u,d,s}{yields}Ae{nu}{sub e} are also presented.
Yu Nishiyama - One of the best experts on this subject based on the ideXlab platform.
-
Model-based kernel Sum Rule: kernel Bayesian inference with probabilistic models
Machine Learning, 2020Co-Authors: Yu Nishiyama, Motonobu Kanagawa, Arthur Gretton, Kenji FukumizuAbstract:Kernel Bayesian inference is a principled approach to nonparametric inference in probabilistic graphical models, where probabilistic relationships between variables are learned from data in a nonparametric manner. Various algorithms of kernel Bayesian inference have been developed by combining kernelized basic probabilistic operations such as the kernel Sum Rule and kernel Bayes’ Rule. However, the current framework is fully nonparametric, and it does not allow a user to flexibly combine nonparametric and model-based inferences. This is inefficient when there are good probabilistic models (or simulation models) available for some parts of a graphical model; this is in particular true in scientific fields where “models” are the central topic of study. Our contribution in this paper is to introduce a novel approach, termed the model-based kernel Sum Rule (Mb-KSR), to combine a probabilistic model and kernel Bayesian inference. By combining the Mb-KSR with the existing kernelized probabilistic Rules, one can develop various algorithms for hybrid (i.e., nonparametric and model-based) inferences. As an illustrative example, we consider Bayesian filtering in a state space model, where typically there exists an accurate probabilistic model for the state transition process. We propose a novel filtering method that combines model-based inference for the state transition process and data-driven, nonparametric inference for the observation generating process. We empirically validate our approach with synthetic and real-data experiments, the latter being the problem of vision-based mobile robot localization in robotics, which illustrates the effectiveness of the proposed hybrid approach.
-
model based kernel Sum Rule kernel bayesian inference with probabilistic models
arXiv: Machine Learning, 2014Co-Authors: Yu Nishiyama, Motonobu Kanagawa, Arthur Gretton, Kenji FukumizuAbstract:Kernel Bayesian inference is a principled approach to nonparametric inference in probabilistic graphical models, where probabilistic relationships between variables are learned from data in a nonparametric manner. Various algorithms of kernel Bayesian inference have been developed by combining kernelized basic probabilistic operations such as the kernel Sum Rule and kernel Bayes' Rule. However, the current framework is fully nonparametric, and it does not allow a user to flexibly combine nonparametric and model-based inferences. This is inefficient when there are good probabilistic models (or simulation models) available for some parts of a graphical model; this is in particular true in scientific fields where "models" are the central topic of study. Our contribution in this paper is to introduce a novel approach, termed the {\em model-based kernel Sum Rule} (Mb-KSR), to combine a probabilistic model and kernel Bayesian inference. By combining the Mb-KSR with the existing kernelized probabilistic Rules, one can develop various algorithms for hybrid (i.e., nonparametric and model-based) inferences. As an illustrative example, we consider Bayesian filtering in a state space model, where typically there exists an accurate probabilistic model for the state transition process. We propose a novel filtering method that combines model-based inference for the state transition process and data-driven, nonparametric inference for the observation generating process. We empirically validate our approach with synthetic and real-data experiments, the latter being the problem of vision-based mobile robot localization in robotics, which illustrates the effectiveness of the proposed hybrid approach.
Arthur J Freeman - One of the best experts on this subject based on the ideXlab platform.
-
limitation of the magnetic circular dichroism spin Sum Rule for transition metals and importance of the magnetic dipole term
Physical Review Letters, 1994Co-Authors: Arthur J FreemanAbstract:Magnetic-circular-dichroism (MCD) spectra and the spin magnetic dipole term ([l angle][ital T][sub [ital z]][r angle]) for bulk 3[ital d] transition metals and their surfaces were calculated from full potential linearized augmented plane wave electronic band structure results. The recently proposed MCD spin Sum Rule is found to result in a much larger error [of up to 50% for the Ni(001) surface] than does its orbital counterpart. In support of recent experiments for bulk, we find that by combining the MCD orbital and spin Sum Rules the ratio of spin and orbital moments can be determined from the MCD spectra even for low dimension systems with an error of 10% when the [l angle][ital T][sub [ital z]][r angle] contribution is included.
-
first principles investigation of the validity and range of applicability of the x ray magnetic circular dichroism Sum Rule
Physical Review Letters, 1993Co-Authors: Dingsheng Wang, Arthur J FreemanAbstract:X-ray magnetic circular dichroism (MCD) spectra and orbital angular momentum, [l angle][ital L][sub [ital z]][r angle], for transition metal bulk and surfaces were studied for both ground state and core hole excitations using a highly precise local density approach. For Fe(001), we predict a double peak structure in both the MCD and total absorption spectra and a strong enhancement of [l angle][ital I][sub [ital z]][r angle]. Surprisingly, the MCD orbital Sum Rule is found to be valid to within (5--10)%. Finally, the results suggest possible solutions to several problems faced in applying the MCD Sum Rule to measure [l angle][ital L][sub [ital z]][r angle].
H. Zaraket - One of the best experts on this subject based on the ideXlab platform.
-
a simple Sum Rule for the thermal gluon spectral function and applications
Journal of High Energy Physics, 2002Co-Authors: P. Aurenche, F. Gelis, H. ZaraketAbstract:In this paper, we derive a simple Sum Rule satisfied by the gluon spectral function at finite temperature. This Sum Rule is useful in order to calculate exactly some integrals that appear frequently in the photon or dilepton production rate by a quark gluon plasma. Using this Sum Rule, we rederive simply some known results and obtain some new results that would be extremely difficult to justify otherwise. In particular, we explain how this result can be used to calculate the photon rate in a simple quasi-particle model which correctly reproduces the thermodynamic properties of a quark-gluon plasma. We also derive an exact expression for the collision integral that appears in the calculation of the Landau-Pomeranchuk-Migdal effect and propose a method to solve the resulting differential equation.
-
A simple Sum Rule for the thermal gluon spectral function and applications
Journal of High Energy Physics, 2002Co-Authors: P. Aurenche, F. Gelis, H. ZaraketAbstract:In this paper, we derive a simple Sum Rule satisfied by the gluon spectral function at finite temperature. This Sum Rule is useful in order to calculate exactly some integrals that appear frequently in the photon or dilepton production rate by a quark gluon plasma. Using this Sum Rule, we rederive simply some known results and obtain some new results that would be extremely difficult to justify otherwise. In particular, we derive an exact expression for the collision integral that appears in the calculation of the Landau-Pomeranchuk-Migdal effect.
