Kernel Partial Least Squares for Stationary Data

Authors: Marco Singer, Tatyana Krivobokova, Axel Munk

JMLR 2017 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental It is shown both theoretically and in simulations that long range dependence results in slower convergence rates. A protein dynamics example shows high predictive power of kernel partial least squares. [...] To validate the theoretical results of the previous sections, we conducted a simulation study.
Researcher Affiliation Academia Marco Singer EMAIL Institute for Mathematical Stochastics Georg-August-Universit at G ottingen, 37077, Germany; Tatyana Krivobokova EMAIL Institute for Mathematical Stochastics Georg-August-Universit at G ottingen, 37077, Germany; Axel Munk EMAIL Institute for Mathematical Stochastics Georg-August-Universit at G ottingen, 37077, Germany
Pseudocode No The paper describes algorithms mathematically and refers to them (e.g., KPLS, KCG) but does not provide structured pseudocode blocks or algorithms.
Open Source Code No The paper does not contain any explicit statement about releasing source code for the described methodology, nor does it provide a link to a code repository.
Open Datasets No The paper mentions using a "protein dynamics example" and "T4 Lysozyme (T4L)" data for its application. While these are specific types of data, the paper does not provide concrete access information (e.g., a link, DOI, or specific repository) for the dataset used in their experiments. It describes the data source as "molecular dynamics simulations" but no public access is indicated.
Dataset Splits Yes The first 50% of the data form a training set to calculate the kernel partial least squares estimator and the remaining data are used for testing.
Hardware Specification No The paper does not contain any specific details about the hardware (e.g., CPU, GPU models) used to run the simulations or experiments.
Software Dependencies No The paper discusses various algorithms and mathematical frameworks but does not specify any software names with version numbers that were used for implementation (e.g., programming languages, libraries, solvers).
Experiment Setup Yes The reproducing kernel Hilbert space is chosen to correspond to the Gaussian kernel k(x, y) = exp( l x y 2), x, y Rd, l = 2, for d = 1. [...] The parameter l > 0 is calculated via cross validation on the training set. In our evaluation we obtained l = 10.22. [...] a maximum of 40 iteration steps. [...] The source parameter is taken to be r = 4.5 and we consider the function f(x) = 4.37 1{3L4(x, 4) 2L4(x, 3) + 1.5L4(x, 9)}, x R. [...] The residuals ε(j) 1 , . . . , ε(j) n are generated as independent standard normally distributed random variables and independent of {X(j) t }n t=1 . The response is defined as y(j) t = f(X(j) t )+ η ε(j) t , t = 1, . . . , n, j = 1, . . . , M, with η = 1/16.