Reservoir Labs Presenting at IEEE High Performance Extreme Computing Conference (HPEC ’18)
Reservoir Labs is presenting three papers at the 2018 IEEE High Performance Extreme Computing Conference (HPEC ’18) in Waltham, MA. Best Paper Finalist “Computationally Efficient CP Tensor Decomposition Update Framework for Emerging Component Discovery in Streaming Data,” shows new ENSIGN® capabilities for large scale streaming sparse tensor or hypergraph decompositions on cyber security log data, including the ability to discover new behaviors occurring on the network through the ability of the numerical algorithm and implementation to form new components. This work adds to Reservoir’s differentiated capabilities in streaming statistical analysis of large scale graph and hypergraph data, utilizing parallel computing and proprietary data structures and algorithms.
Reservoir Labs is also presenting “All-at-once Decomposition of Coupled Billion-scale Tensors in Apache Spark,” which shows how to perform very large scale hypergraph decomposition and sparse tensor decomposition utilizing the Apache Spark platform, including the ability to perform this for coupled sparse tensor decomposition. The results surface meaningful and interesting behaviors in cyber security and geospatial activity databases. A third paper, “Accelerating Dijkstra’s Algorithm Using Multiresolution Priority Queues,” presents Reservoir’s research in achieving an asymptotic improvement in shortest path finding for very large scale graph data analytics through the application of Reservoir’s proprietary multiresolution queue data structure.