A sampling-based verification framework for AI inference that replaces full cryptographic proofs with Merkle-tree commitments, reducing proving times from minutes to milliseconds …
We present a method for detecting structural events in log message streams. The approach models log data to identify event patterns without requiring prior domain knowledge, …
We develop a novel algorithm for detecting hotspots in service-oriented architectures. By mining metrics associated with service calls, the method recommends services that are the …
We develop parallel versions of graph mining algorithms that achieve significant improvement in runtime. The approach exploits GPU parallelism to accelerate frequent subgraph …
We present novel and scalable methods for approximate frequent subgraph mining from exact and probabilistic graphs. By incorporating label costs, the approach yields more …
We develop a spectral approach for detecting communities in signed networks, where edges carry positive and negative weights. The method leverages structural balance theory to find …
We apply large sparse graph mining to discover infrastructure patterns in configuration management databases (CMDBs). The method identifies recurring structural patterns in IT …