Name: Kyle Kloster
Type: User
Company: Data science, scientific computing
Bio: Data, algorithms, graphs, and matrices. SWE at Carbon3d doing Computational Geometry. Former Math/CS PhD research at Purdue Univ and NCSU.
Location: San Francisco Bay
Blog: https://kkloste.github.io/
Kyle Kloster's Projects
Manuscript describing Hetnet Connectivity Search
Variety of local graph diffusion codes for targeted network analysis and community detection
Heat-kernel based local clustering
Toward scalable, robust community detection and other semi-supervised learning tasks
Algorithms for enumerating maximal induced bicliques. OCT-MIB is designed for graphs which are "near bipartite" described in Kloster et al. 2018+; LexMIB is an improved version of the algorithm described in Dias et al. 2005.
Matrix Exponential for node centrality, link-prediction, and more
Local graph analysis via PageRank solution paths
HTML interface for querying competitive Smash 64 tournament archive
A practical fpt algorithm for Flow Decomposition and transcript assembly
Manuscript on XSwap network permutation and hetnet node degrees