An $\alpha$-No-Regret Algorithm For Graphical Bilinear Bandits

Authors: Geovani Rizk, Igor Colin, Albert Thomas, Rida Laraki, Yann Chevaleyre

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, we show through various experiments the validity of our approach. and Section 5 gives an experimental overview of the various theoretical bounds established for the regret.
Researcher Affiliation Collaboration Geovani Rizk PSL Université Paris Dauphine, CNRS, LAMSADE, Paris, France... Igor Colin Huawei Noah s Ark Lab, Paris, France... Albert Thomas Huawei Noah s Ark Lab, Paris, France... Rida Laraki PSL Université Paris Dauphine, CNRS, LAMSADE, Paris, France... Yann Chevaleyre PSL Université Paris Dauphine, CNRS, LAMSADE, Paris, France
Pseudocode Yes Algorithm 1: Adaptation of OFUL algorithm for Graphical Bilinear Bandit, Algorithm 2: Approx-MAX-CUT, Algorithm 3: Improved OFUL for Graphical Bilinear Bandits
Open Source Code No Did you include the code, data, and instructions needed to reproduce the main experimental results (either in the supplemental material or as a URL)? [No] The code will be released if the paper is accepted to the conference
Open Datasets No No concrete access information (specific link, DOI, repository name, formal citation with authors/year, or reference to established benchmark datasets) for a publicly available or open dataset was found. The paper mentions "synthetic datasets" but provides no access details.
Dataset Splits No No specific dataset split information (exact percentages, sample counts, citations to predefined splits, or detailed splitting methodology) needed to reproduce the data partitioning was found.
Hardware Specification No Did you include the total amount of compute and the type of resources used (e.g., type of GPUs, internal cluster, or cloud provider)? [Yes] Also in the appendix. (The main paper text provided does not contain these details.)
Software Dependencies No No specific ancillary software details (e.g., library or solver names with version numbers like Python 3.8, CPLEX 12.4) needed to replicate the experiment were found in the main paper.
Experiment Setup Yes Experiments were performed on graphs of n = 100 nodes, and results for the random graph are averaged over 100 draws... The dimension d of the arm-set is 10 (which gives linear reward with unknown parameter ? of dimension 100). The plotted curve represents the average value of the parameters over 100 different matrices M? initiated randomly with positive values... We use a complete graph of 5 nodes, we run the experiment on 5 different matrices as in Figure 1 with = 0 and run it 10 different times to plot the average fraction of the global reward.