The Multiscale Laplacian Graph Kernel

Authors: Risi Kondor, Horace Pan

NeurIPS 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We tested the efficacy of the MLG kernel by performing classification on benchmark bioinformatics datasets using a binary C-SVM solver [21], and compared our classification results against those from other representative graph kernels from the literature: the Weisfeiler Lehman Kernel, the Weisfeiler Lehman Edge Kernel [9], the Shortest Path Kernel [6], the Graphlet Kernel [9], and the p-random Walk Kernel [5].
Researcher Affiliation Academia Risi Kondor Department of Computer Science Department of Statistics University of Chicago Chicago, IL 60637 risi@cs.uchicago.edu Horace Pan Department of Computer Science University of Chicago Chicago, IL 60637 hopan@uchicago.edu
Pseudocode No The paper describes algorithms and procedures in prose and mathematical formulations but does not include any explicitly labeled pseudocode or algorithm blocks.
Open Source Code No The paper does not contain any explicit statements about releasing source code or provide links to a code repository for the described methodology.
Open Datasets Yes We tested the efficacy of the MLG kernel by performing classification on benchmark bioinformatics datasets...MUTAG[22] PTC[23] ENZYMES[2] PROTEINS[2] NCI1[24] NCI109[24]
Dataset Splits Yes On the other 80% we did 10 fold cross validation to select the parameters for each kernel method to be used on the test set and repeated this setup 10 times.
Hardware Specification Yes All experiments were done on a 16 core Intel E5-2670 @ 2.6GHz processor with 32 GB of memory.
Software Dependencies No The paper mentions using a "binary C-SVM solver [21]" (referring to LibSVM), but does not provide specific version numbers for any software dependencies.
Experiment Setup Yes For the Weisfeiler Lehman kernels, the height parameter h is chosen from {1, 2, ..., 5}, the random walk size p for the p-random walk kernel was chosen from {1, 2, ..., 5}, for the Graphlets kernel the graphlet size n was chosen from {3, 4, 5}. For the parameters of the MLG kernel: we chose η from {0.01, 0.1, 1}, radius size n from {1, 2, 3}, number of levels l from {1, 2, 3}, and fixed gamma to be 0.01.