Maximum Likelihood Embedding of Logistic Random Dot Product Graphs

Authors: Luke J. O'Connor, Muriel Medard, Soheil Feizi5289-5297

AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In simulations, we show that our proposed method is more accurate and more robust than common practices. We also show the effectiveness of our approach over standard real networks of the karate club and political blogs.
Researcher Affiliation Academia Luke J. O Connor Broad Institute of MIT and Harvard 415 Main St, Cambridge, MA 02142 loconnor@broadinstitute.org Muriel Medard Massachusetts Institute of Technology 77 Massachusetts Ave, Cambridge, MA 02139 medard@mit.edu Soheil Feizi University of Maryland 8125 Paint Branch Drive, College Park, MD 20742 sfeizi@cs.umd.edu
Pseudocode Yes Input: Adjacency matrix A, number of dimensions d. (optional) number of clusters k Step 0: Form the mean-centered adjacency matrix B := A 1/(n(n 1)) A . Step 1: Compute d eigenvectors of B with largest eigenvalues: e1, ..., ed. Step 2: Let Xi = eie T i for 1 i d. Perform logistic regression of the entries of A lying above the diagonal on the corresponding entries of X1, ...., Xd, estimating coefficients λ 1, ...λ d subject to the constraint that λi 0 i. Output: Let V be the matrix formed by concatenating λ 1e1, ..., λ ded. Return V . (optional): Perform k-means on V , and return the inferred clusters.
Open Source Code No The paper does not provide explicit statements about open-source code availability or links to code repositories.
Open Datasets Yes First, we consider the Karate Club network (Zachary 1977). In the second experiment, we consider a network of political blogs, whose edges correspond to links between blogs (Adamic and Glance 2005).
Dataset Splits No The paper discusses synthetic network generation for performance evaluation but does not provide specific training, validation, and test dataset splits for its experiments.
Hardware Specification No The paper does not specify the hardware (e.g., GPU/CPU models, memory) used to run the experiments.
Software Dependencies No The paper does not provide specific software dependencies with version numbers.
Experiment Setup No The paper describes the steps of its proposed algorithm but does not provide specific experimental setup details such as hyperparameter values, optimizer settings, or detailed training configurations.