Central and Noncentral Limit Theorems Arising from the Scattering Transform and Its Neural Activation Generalization

DOI10.1137/21M1454511MaRDI QIDQ6102371zbMATH OpenOpenAlexWikidataFDO

Authors Gi-Ren Liu, Yuan-Chung Sheu, H.-T. Wu

Publication date 8 May 2023

Published in SIAM Journal on Mathematical Analysis (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/2011.10801

zbMATH Keywords

deformation insensitivedependent processeslarge scaling limitslong range dependenceneural activation scattering transformscattering momentstranslation invariant

Mathematics Subject Classification ID

General harmonic expansions, frames (42C15) Nontrigonometric harmonic analysis involving wavelets and other special systems (42C40) Central limit and other weak theorems (60F05) Random fields (60G60) Inference from stochastic processes and spectral analysis (62M15)

Abstract: Motivated by analyzing complicated and non-stationary time series, we study a generalization of the scattering transform (ST) that includes broad neural activation functions, which is called neural activation ST (NAST). On the whole, NAST is a transform that comprises a sequence of ``neural processing units, each of which applies a high pass filter to the input from the previous layer followed by a composition with a nonlinear function as the output to the next neuron. Here, the nonlinear function models how a neuron gets excited by the input signal. In addition to showing properties like non-expansion, horizontal translational invariability and insensitivity to local deformation, the statistical properties of the second order NAST of a Gaussian process with various dependence and (non-)stationarity structure and its interaction with the chosen high pass filters and activation functions are explored and central limit theorem (CLT) and non-CLT results are provided. Numerical simulations are also provided. The results explain how NAST processes complicated and non-stationary time series, and pave a way towards statistical inference based on NAST under the non-null case.

Recommendations

• Asymptotic analysis of higher-order scattering transform of Gaussian processes
• The Scattering Transform
• Analysis of time-frequency scattering transforms
• Group invariant scattering
• A uniform central limit theorem for neural network-based autoregressive processes with applications to change-point analysis

Cites work

• A Mathematical Theory of Deep Convolutional Neural Networks for Feature Extraction
• Analysis of time-frequency scattering transforms
• Asymptotic analysis of higher-order scattering transform of Gaussian processes
• Asymptotic theory of statistics and probability
• Central limit theorems for non-linear functionals of Gaussian fields
• Convergence of integrated processes of arbitrary Hermite rank
• Deep Haar scattering networks
• Deep Scattering Spectrum
• Explore Intrinsic Geometry of Sleep Dynamics and Predict Sleep Stage by Unsupervised Learning Techniques
• Fractional Brownian Motions, Fractional Noises and Applications
• Generalized Morse wavelets
• Graph convolutional neural networks via scattering
• Group invariant scattering
• Higher-order spectral densities of fractional random fields
• scientific article; zbMATH DE number 1715060 (Why is no real title available?)
• scientific article; zbMATH DE number 43570 (Why is no real title available?)
• scientific article; zbMATH DE number 1284881 (Why is no real title available?)
• scientific article; zbMATH DE number 3274494 (Why is no real title available?)
• scientific article; zbMATH DE number 3331075 (Why is no real title available?)
• Intermittent process analysis with scattering moments
• Large scale behavior of wavelet coefficients of non-linear subordinated processes with long memory
• Limit theorems for weighted nonlinear transformations of Gaussian stationary processes with singular spectra
• Long-Range Dependence and Self-Similarity
• Maximum entropy models from phase harmonic covariances
• Modelling long-range-dependent Gaussian processes with application in continuous-time financial models
• Multiple Wiener-Ito integrals. With applications to limit theorems
• Multiscale sparse microcanonical models
• Non-central limit theorems for non-linear functional of Gaussian fields
• Normal approximations with Malliavin calculus. From Stein's method to universality
• Phase harmonic correlations and convolutional neural networks
• Regularization and integral representations of Hermite processes
• Regularly varying functions
• Rotationally invariant time-frequency scattering transforms
• Scaling limits of solutions of the heat equation for singular non-Gaussian data
• Spectral representation of fractional Brownian motion in n dimensions and its properties
• Ten Lectures on Wavelets
• Wavelet estimator of long-range dependent processes.
• Wavelet Scattering Regression of Quantum Chemical Energies
• Wiener integrals with respect to the two-parameter tempered Hermite random fields

Cited in

(6)

• Asymptotic analysis of higher-order scattering transform of Gaussian processes
• Analysis of time-frequency scattering transforms
• The Scattering Transform
• When scattering transform meets non-Gaussian random processes, a double scaling limit result
• Convergence rate analysis in limit theorems for nonlinear functionals of the second Wiener chaos
• Gaussian approximation for the moving averaged modulus wavelet transform and its variants

This page was built for publication: Central and Noncentral Limit Theorems Arising from the Scattering Transform and Its Neural Activation Generalization

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6102371)