Online Influence Maximization under Independent Cascade Model with Semi-Bandit Feedback

Authors: Zheng Wen, Branislav Kveton, Michal Valko, Sharan Vaswani

NeurIPS 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experiments show that in several representative graph topologies, the regret of IMLin UCB scales as suggested by our upper bounds. Our experiments also show that IMLin UCB with linear generalization can lead to low regret in real-world online influence maximization.
Researcher Affiliation Collaboration Zheng Wen Adobe Research zwen@adobe.com Branislav Kveton Adobe Research kveton@adobe.com Michal Valko Seque L team, INRIA Lille Nord Europe michal.valko@inria.fr Sharan Vaswani University of British Columbia sharanv@cs.ubc.ca
Pseudocode Yes Algorithm 1 IMLin UCB: Influence Maximization Linear UCB
Open Source Code No The paper does not provide any explicit statements about releasing source code or links to a code repository for the methodology described.
Open Datasets Yes Specifically, we compare IMLin UCB with CUCB in a subgraph of Facebook network from [22]. [22] Jure Leskovec and Andrej Krevl. Snap datasets: Stanford large network dataset collection. http://snap.stanford.edu/data, jun 2014.
Dataset Splits No The paper mentions using a 'subgraph of Facebook network' but does not specify exact train/validation/test dataset splits (e.g., percentages or counts) or reference standard predefined splits with specific details for reproducibility.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running its experiments.
Software Dependencies No The paper mentions using 'node2vec algorithm [15]' and 'offline IM algorithm proposed in [27]' but does not provide specific version numbers for these or any other software dependencies needed for replication.
Experiment Setup Yes We set n = 5000 and K = 10 in this experiment. For IMLin UCB, we choose d = 10 and generate edge feature xe s as follows: we first use node2vec algorithm [15] to generate a node feature in ℜd for each node v V; then for each edge e, we generate xe as the element-wise product of node features of the two nodes connected to e. ... For both CUCB and IMLin UCB, we choose ORACLE as the state-of-the-art offline IM algorithm proposed in [27].