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..
Provably Efficient Multi-Task Meta Bandit Learning via Shared Representations
Authors: Jiabin Lin, Shana Moothedath
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We validated the theoretical results through numerical experiments using real-world and synthetic datasets, comparing them against benchmark algorithms. |
| Researcher Affiliation | Academia | Jiabin Lin and Shana Moothedath Department of Electrical and Computer Engineering Iowa State University EMAIL, EMAIL |
| Pseudocode | Yes | Algorithm 1 Explore-then-Commit (Et C) Algorithm for Representation Learning in Linear Bandits Algorithm 2 OFUL-Based Meta Bandit Learning using Shared Representations |
| Open Source Code | Yes | Question: Does the paper provide open access to the data and code, with sufficient instructions to faithfully reproduce the main experimental results, as described in supplemental material? Answer: [Yes] Justification: We provide open access to the data and code, along with thorough documentation, enabling others to precisely reproduce the experimental results. (https: //github.com/somethingputhere/Simulation.git) |
| Open Datasets | Yes | We utilized the Movielens-100K dataset [52], which contains user ratings for movies. The Last FM dataset is from the online music streaming service Last.fm, including data for 1892 users and 17632 artists. |
| Dataset Splits | No | The paper describes how synthetic data is generated and how Movielens and Last FM datasets are pre-processed for use in experiments (e.g., rating normalization, non-negative matrix factorization). However, it does not explicitly provide information on training, validation, or test dataset splits (e.g., percentages, sample counts, or references to standard splits). |
| Hardware Specification | No | Question: For each experiment, does the paper provide sufficient information on the computer resources (type of compute workers, memory, time of execution) needed to reproduce the experiments? Answer: [No] Justification: We omitted detailed information regarding the computational resources utilized, as our concentration is mainly on theoretical analysis (e.g., cumulative regret and sample complexity), and the experiments are computationally lightweight. |
| Software Dependencies | No | The paper does not explicitly list specific software dependencies with version numbers (e.g., Python 3.8, PyTorch 1.9, CUDA 11.1) in the main text or the provided NeurIPS checklist. |
| Experiment Setup | Yes | Synthetic data: The B Rd r matrix is generated by orthonormalizing an i.i.d. standard Gaussian matrix, and W Rr T is generated from an i.i.d. Gaussian distribution. The feature matrices ÎŚn,ts are generated from the standard Gaussian distribution. We set d = 100, T = 100, and r = 2, and alter the parameters d, T, and r in the experiments to assess performance. In the transfer learning setting, we utilize d = 100, T = 200, and r = 2. All results are averaged over 100 independent trials. The error bars show standard errors, calculated as standard deviations divided by. Movielens: ...We set parameters as: d = 100, T = 10, and r = 1. Last FM: ...We set parameters as: d = 100, T = 10, and r = 1. Figure 1: Results of representation learning: ...for d = 100, T = 100, N1 = 45, and N = 200. ...The parameters are set as d = 100, T = 10, r = 1, N1 = 300, N = 1000. Figure 2: Results of transfer learning: ...for d = 100, T = 200, r = 2, N1 = 400. ...for d = 100, T = 10, r = 1, N1 = 1000. ...for d = 100, T = 10, r = 1, N1 = 500. |