Streaming Robust Submodular Maximization: A Partitioned Thresholding Approach

Authors: Slobodan Mitrovic, Ilija Bogunovic, Ashkan Norouzi-Fard, Jakub M. Tarnawski, Volkan Cevher

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

Reproducibility Variable Result LLM Response
Research Type Experimental In two different data summarization tasks, we demonstrate that it matches or outperforms existing greedy and streaming methods, even if they are allowed the benefit of knowing the removed subset in advance.
Researcher Affiliation Academia Slobodan Mitrovi c EPFL Ilija Bogunovic EPFL Ashkan Norouzi-Fard EPFL Jakub Tarnawski EPFL Volkan Cevher EPFL e-mail: slobodan.mitrovic@epfl.ch e-mail: ilija.bogunovic@epfl.ch e-mail: ashkan.norouzifard@epfl.ch e-mail: jakub.tarnawski@epfl.ch e-mail: volkan.cevher@epfl.ch
Pseudocode Yes Algorithm 1 STre Aming Robust Thresholding submodular algorithm (STAR-T) and Algorithm 2 STAR-TGREEDY
Open Source Code No No explicit statement or link for open-source code release is provided.
Open Datasets Yes We consider two datasets: (i) ego-Twitter [19], consisting of 973 social circles from Twitter, which form a directed graph with 81306 nodes and 1768149 edges; (ii) Amazon product co-purchasing network [20]: a directed graph with 317914 nodes and 1745870 edges. We use the Movie Lens 1M database [21], which contains 1000209 ratings for 3900 movies by 6040 users.
Dataset Splits No No explicit information on training, validation, or test dataset splits (e.g., percentages, counts, or cross-validation folds) is provided for reproducibility of data partitioning.
Hardware Specification No No specific hardware (e.g., GPU/CPU models, memory) used for experiments is mentioned in the paper.
Software Dependencies No No specific software dependencies with version numbers are mentioned in the paper.
Experiment Setup Yes As the input, STAR-T requires a non-negative monotone submodular function f, cardinality constraint k, robustness parameter m and thresholding parameter τ. For a user u, we use the following monotone submodular function to recommend a set of movies Z: fu(Z) = (1 α) X z Z vu, vz + α X m M max z Z vm, vz .