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 [1].
A Wasserstein Distance Approach for Concentration of Empirical Risk Estimates
Authors: Prashanth L.A., Sanjay P. Bhat
JMLR 2022 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Theoretical | This paper presents a unified approach based on Wasserstein distance to derive concentration bounds for empirical estimates for two broad classes of risk measures defined in the paper. The bounds are derived by first relating the estimation error to the Wasserstein distance between the true and empirical distributions, and then using recent concentration bounds for the latter. In comparison to the conference version, this paper includes new theoretical results, all the convergence proofs, and a revised presentation. The usefulness of the bounds is illustrated through an algorithm and the corresponding regret bound for a stochastic bandit problem... |
| Researcher Affiliation | Collaboration | Prashanth L.A. EMAIL Department of Computer Science and Engineering Indian Institute of Technology Madras, Chennai 600036, India Sanjay P. Bhat EMAIL TCS Research Hyderabad 500081, India |
| Pseudocode | Yes | Risk-LCB algorithm Initialization: Play each arm once. For t = K + 1, . . . , n, repeat 1. For each arm i = 1, . . . , K, define LCBt(i) = ρi,Ti(t 1) wi,Ti(t 1), where ρi,Ti(t 1) is the estimate of the risk measure for arm i computed using (5) from Ti(t 1) samples, and wi,Ti(t 1) is defined in (33). 2. Play arm It = arg min i=1,...,K LCBt(i). 3. Observe sample Xt from the distribution PIt corresponding to the arm It. |
| Open Source Code | No | The paper does not provide an explicit statement about the release of source code for the methodology described, nor does it include a link to a code repository. The license mentioned refers to the paper's publication, not its implementation code. |
| Open Datasets | No | The paper is theoretical in nature, focusing on deriving concentration bounds for risk measures under various distribution assumptions (sub-Gaussian, sub-exponential, heavy-tailed distributions). It does not describe experiments performed on specific datasets or provide any access information for such. |
| Dataset Splits | No | The paper is theoretical and does not involve empirical experiments using specific datasets, therefore, no dataset split information is provided. |
| Hardware Specification | No | The paper is theoretical, presenting mathematical derivations, proofs, and algorithms. it does not describe any empirical experiments or the hardware used to run them. |
| Software Dependencies | No | The paper is theoretical and does not describe an implementation or empirical evaluation that would require specific software dependencies with version numbers. |
| Experiment Setup | No | The paper is primarily theoretical, dealing with concentration bounds and risk-sensitive bandit algorithms. It does not include an empirical experimental setup with details such as hyperparameters or training configurations. |