Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Kernel von Mises Formula of the Influence Function
Authors: Yaroslav Mukhin
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The Appendix contains simulation experiments and deferred technical details. Example 4.3... We simulate Monte Carlo data from the standard Normal distribution, corresponding to the shape parameter α = 1/2 of our Example 4.3. This allows us to compute the oracle ψr λ using Hermit polynomials that we numerically evaluate using the MATLAB code provided with the textbook [FM15]. We estimate the eigenvalues σj and eigenfunctions ej(Xi) the using Nyström method via MATLAB s eig function. We estimate the pathwise derivatives as 1 n Pn i=1 Xiˆej(Xi), note this does not take into account estimation of the density and evaluation of the mean functional on the estimated distribution. Figure 3: Surrogate bias ψ ψr λ 2 L2(P ), mean integrated squared error E ψr λ ˆψr λ 2 L2(P ), and distribution of the error ψr λ ˆψr λ L2(P ) based on 103 Monte Carlo experiments. |
| Researcher Affiliation | Academia | Yaroslav Mukhin Cornell University EMAIL |
| Pseudocode | No | The paper describes mathematical derivations and theoretical concepts, but it does not contain any clearly labeled pseudocode or algorithm blocks with structured steps. |
| Open Source Code | Yes | Justification: We provide our code for the toy Monte Carlo experiment. |
| Open Datasets | No | We simulate Monte Carlo data from the standard Normal distribution, corresponding to the shape parameter α = 1/2 of our Example 4.3. |
| Dataset Splits | No | The paper mentions sample sizes used in the Monte Carlo experiments (e.g., n=64, n=256, n=1024) but does not define training, validation, or test splits for a pre-existing dataset. |
| Hardware Specification | Yes | Justification: Apple M2 mac. |
| Software Dependencies | No | We numerically evaluate using the MATLAB code provided with the textbook [FM15]. We estimate the eigenvalues σj and eigenfunctions ej(Xi) the using Nyström method via MATLAB s eig function. The paper mentions MATLAB and MATLAB code but does not provide specific version numbers for MATLAB or any other software dependencies. |
| Experiment Setup | Yes | We use the Gaussian PSD kernel from our Example 4.3 and set the shape parameter ϵ = 1. We simulate Monte Carlo data from the standard Normal distribution, corresponding to the shape parameter α = 1/2 of our Example 4.3. For a fixed rank 1 r n and regularization loading λ 0, let DθP [ej] 1 + 2λ/σj ej(x), ˆψr λ(x) := 1 1 + 2λ/ˆσj dtθ( ˆf j t ) |t=0 ˆej(x) |