Contextual Multinomial Logit Bandits with General Value Functions

Authors: Mengxiao Zhang, Haipeng Luo

NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical In this work, we consider contextual MNL bandits with a general value function class that contains the ground truth, borrowing ideas from a recent trend of studies on contextual bandits. Specifically, we consider both the stochastic and the adversarial settings, and propose a suite of algorithms, each with different computation-regret trade-off. ... Throughout the paper, we use two running examples to illustrate the concrete regret bounds our different algorithms achieve: the finite class and the linear class. In particular, for the linear class, this leads to five new results, summarized in Table 1 together with previous results.
Researcher Affiliation Academia Mengxiao Zhang University of Iowa mengxiao-zhang@uiowa.edu Haipeng Luo University of Southern California haipengl@usc.edu
Pseudocode Yes Algorithm 1 Contextual MNL Algorithms with an Offline Regression Oracle, Algorithm 2 Contextual MNL Algorithms via an Online Regression Oracle, Algorithm 3 Feel-Good Thompson Sampling for Contextual MNL bandits
Open Source Code No The paper does not contain any statement about releasing source code for the described methodology, nor does it provide any links to a code repository.
Open Datasets No The paper is purely theoretical and does not conduct experiments using publicly available datasets. It refers to 'finite class' and 'linear class' as abstract function classes for theoretical analysis, not concrete datasets.
Dataset Splits No This paper is theoretical and does not involve the use of datasets with training, validation, or test splits for empirical evaluation.
Hardware Specification No This paper is purely theoretical and does not describe any specific hardware (e.g., GPU models, CPU types, memory) used for running experiments.
Software Dependencies No This paper is purely theoretical and does not mention any specific software dependencies with version numbers (e.g., programming languages, libraries, frameworks) required to reproduce experiments.
Experiment Setup No This paper is purely theoretical and does not describe any experimental setup details, such as hyperparameter values, optimization settings, or training configurations.