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..
Parameter-free Algorithms for the Stochastically Extended Adversarial Model
Authors: Shuche Wang, Adarsh Barik, Peng Zhao, Vincent Tan
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Theoretical | We develop the first parameter-free algorithms for the Stochastically Extended Adversarial (SEA) model, a framework that bridges adversarial and stochastic online convex optimization. Existing approaches for the SEA model require prior knowledge of problem-specific parameters, such as the diameter of the domain D and the Lipschitz constant of the loss functions G, which limits their practical applicability. Addressing this, we develop parameter-free methods by leveraging the Optimistic Online Newton Step (OONS) algorithm to eliminate the need for these parameters. We first establish a comparator-adaptive algorithm for the scenario with unknown domain diameter but known Lipschitz constant, achieving an expected regret bound of e O u 2 2 + u 2( p Σ2 1:T ) , where u is the comparator vector and σ2 1:T and Σ2 1:T represent the cumulative stochastic variance and cumulative adversarial variation, respectively. We then extend this to the more general setting where both D and G are unknown, attaining the comparatorand Lipschitz-adaptive algorithm. Notably, the regret bound exhibits the same dependence on σ2 1:T and Σ2 1:T , demonstrating the efficacy of our proposed methods even when both parameters are unknown in the SEA model. The NeurIPS Paper Checklist also explicitly states: "This paper is a fully theoretical paper without an experimental section." |
| Researcher Affiliation | Academia | 1 Institute of Operations Research and Analytics, National University of Singapore 2 Department of Computer Science and Engineering, Indian Institute of Technology Delhi 3 National Key Laboratory for Novel Software Technology, Nanjing University 4 School of Artificial Intelligence, Nanjing University 5 Department of Mathematics, National University of Singapore 6 Department of Electrical and Computer Engineering, National University of Singapore |
| Pseudocode | Yes | Algorithm 1 Optimistic Online Newton Step (OONS) Input: learning rate ηt > 0, x 1 = 0. 1: for t = 1, . . . , T do ... Algorithm 2 Comparator-adaptive algorithm for the SEA model (CA-OONS) Input: Lipschitz constant G. 1: for t = 1, . . . , T do ... Algorithm 3 Meta Algorithm Input: Additional expert set S defined in (7). Initialization: p 1 S such that p 1,k β2 k for all k S. 1: for t = 1, . . . , T do ... Algorithm 4 Comparator and Lipschitz-Adaptive (or CLA-OONS) for the SEA model Input: Initial scale B0. Initialize: D1 = 1. 1: for t = 1, . . . , T do ... |
| Open Source Code | No | The NeurIPS Paper Checklist states: "This paper is a fully theoretical paper without an experimental section." and "The paper does not include experiments requiring code." There is no explicit mention of code release or links to repositories in the paper. |
| Open Datasets | No | The NeurIPS Paper Checklist states: "This paper is a fully theoretical paper without an experimental section." and "The paper does not include experiments requiring code." The paper focuses on theoretical aspects of online convex optimization and does not mention the use of any datasets for experimental evaluation. |
| Dataset Splits | No | The paper does not involve experimental research using datasets, therefore, there is no discussion of dataset splits. |
| Hardware Specification | No | The paper is theoretical and does not conduct experiments, hence no hardware specifications are provided. |
| Software Dependencies | No | The paper is theoretical and does not conduct experiments, hence no software dependencies or versions are specified. |
| Experiment Setup | No | The paper is theoretical and does not conduct experiments, hence no experimental setup details, hyperparameters, or training configurations are provided. |