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..
FraPPE: Fast and Efficient Preference-Based Pure Exploration
Authors: Udvas Das, Apurv Shukla, Debabrota Basu
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Finally, we perform numerical experiments across synthetic and real datasets demonstrating that Fra PPE achieves the lowest sample complexities to identify the exact Pareto set among the existing algorithms. |
| Researcher Affiliation | Academia | Udvas Das Univ. Lille, Inria CNRS, Centrale Lille UMR 9189 CRIStAL, Lille, France EMAIL Apurv Shukla Department of EECS, University of Michigan Ann Arbor, MI, USA EMAIL Debabrota Basu Univ. Lille, Inria CNRS, Centrale Lille UMR 9189 CRIStAL, Lille, France EMAIL |
| Pseudocode | Yes | Algorithm 1 Fra PPEFrugal and Fast Preference-Based Pure Exploration |
| Open Source Code | Yes | The anonymous repository link for the implementations code repository. |
| Open Datasets | Yes | Finally, we perform numerical experiments across synthetic and real datasets demonstrating that Fra PPE achieves the lowest sample complexities to identify the exact Pareto set among the existing algorithms. SNW dataset (K = 206, L = 2) is derived from the domain of computational hardware design, specifically concerning the optimization of sorting network configurations (Zuluaga et al., 2012b). Cov-Boost Trial Dataset. This real-lide inspired data set contains tabulated entries of phase-2 booster trial for Covid-19 (Munro et al., 2021). Cov-Boost has been used as a benchmark dataset for evaluating algorithms for Pareto Set Identification (PSI). |
| Dataset Splits | No | The paper discusses using the Cov-Boost Trial Dataset and a Gaussian instance, describing their characteristics and how they are used as bandit environments. However, it does not specify explicit training, validation, or test dataset splits in the conventional sense (e.g., percentages or sample counts) for these datasets, which is typical for pure exploration problems where the goal is to identify optimal arms through sequential sampling rather than training a predictive model on pre-split data. |
| Hardware Specification | Yes | We run all the algorithms on a 64-bit 13th Gen Intel octa-Core i7-1370P 20 processor machine with 32GB RAM. |
| Software Dependencies | No | The paper mentions several solvers and libraries such as CPLEX, HIGHS, CVXOPT, and CLARABEL, and implies the use of Python 3. However, it does not provide specific version numbers for any of these software components, which is required for reproducible software dependencies. |
| Experiment Setup | Yes | We fix δ = 0.01. We consider c(t, δ) = ln( 1+ln(t) /δ ) as the stopping threshold parameter, similar to (Kone et al., 2024). Correlation coefficients are varied from 1 to 1 with grid size 0.1. Algorithm 1: Input: Confidence level δ and sequence {rt}t 1 = t 0.9/K |