Spectral Modification of Graphs for Improved Spectral Clustering

Authors: Ioannis Koutis, Huong Le

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We provide a MATLAB implentation. We plan to provide a Python implementation in the near future. The submitted code and all future updates can be found in: https://github.com/ikoutis/spectral-modification. We performed experiments with four graphs (Blog Catalog, PPI, Wikipedia, Flickr) used as a benchmark in the recent literature [20, 22]. We compare against Net MF [22] as it has previously reported an improvement over Deep Walk [20] and other competing methods.
Researcher Affiliation Academia Ioannis Koutis Department of Computer Science New Jersey Institute of Technology Newark, NJ 07102 ikoutis@njit.edu; Huong Le Department of Computer Science New Jersey Institute of Technology Newark, NJ 07102 hyl4@njit.edu
Pseudocode Yes 1: procedure ENERGY_TD(G, k) ... 9: return {T1, . . . , Tk}
Open Source Code Yes The submitted code and all future updates can be found in: https://github.com/ikoutis/spectral-modification
Open Datasets Yes We performed experiments with four graphs (Blog Catalog, PPI, Wikipedia, Flickr) used as a benchmark in the recent literature [20, 22].
Dataset Splits Yes Figure 4: Micro-F1 scores in 10x cross-validation using LIBLINEAR [10].
Hardware Specification No The paper mentions that "one eigenvector can be computed in mere seconds on standard hardware" but does not provide specific details (e.g., CPU, GPU models, memory) about the hardware used for their experiments.
Software Dependencies No The paper states "We provide a MATLAB implentation." and mentions using "LIBLINEAR [10]" and comparing against "Net MF [22]". However, it does not specify version numbers for MATLAB or any other software components used in their implementation.
Experiment Setup Yes Parameter Settings: For all our experiments we set k = 3, df = 1/2, and α = 1 in ENERGY_TD.