Errors-in-variables Fr\'echet Regression with Low-rank Covariate Approximation

Authors: Dogyoon Song, Kyunghee Han

NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Furthermore, our numerical experiments provide empirical evidence that supports the theoretical findings, demonstrating the superior performance of our approach.
Researcher Affiliation Academia Dogyoon Song Department of EECS University of Michigan Ann Arbor, MI 48109, USA dogyoons@umich.edu Kyunghee Han Department of Math, Stat and Comp Sci University of Illinois at Chicago Chicago, IL 60607, USA hankh@uic.edu
Pseudocode No The paper does not contain any pseudocode or clearly labeled algorithm blocks.
Open Source Code No The paper does not contain any explicit statements or links indicating that the source code for the described methodology is publicly available.
Open Datasets No The paper states that the datasets are 'generated as follows' and provides the methodology for generating synthetic data, but it does not provide access (e.g., a link, DOI, or citation to a repository) to specific pre-generated datasets used in the experiments.
Dataset Splits No The paper mentions a 'training set Dn' and a 'test set Dnew N' but does not explicitly describe a separate validation dataset split.
Hardware Specification No The paper does not provide any specific details about the hardware used to run the experiments.
Software Dependencies No The paper does not specify any software dependencies with version numbers.
Experiment Setup Yes Experimental setup. We consider combinations of p {150, 300, 600} and n {100, 200, 400}. The datasets Dn = {(Xi, Yi) : i [n]} and e Dn = {(Zi, Yi) : i [n]} are generated as follows. Let Xi Np 0p, Σ be IID multivariate Gaussian with mean 0p and covariance Σ such that spec (Σ) = {κj > 0 : j [p]} is an exponentially decreasing sequence... We generate Zi following (5) under two scenarios εij IID N 0, σ2 ε and εij IID Laplace 0, σε , respectively. Lastly, given X = x, let Y be the distribution function of N µα,β(x)+η, τ 2 , where (i) µα,β(x) = α + β x with α = 1 and β = p 1/2 1p; (ii) η N 0, σ2 η ; and (iii) τ 2 IG(s1, s2), an inverse gamma distribution with shape s1 and scale s2. We set σ2 ε = 0.052. We set σ2 η = 0.52, and (s1, s2) = (18, 17). Tuning parameter λ) For simplicity, we chose a universal threshold value as ˆλn = arg minλ Λ MSPE(φ(λ) e Dn), where Λ is a fine grid on 0, pλ1 p/n .