Sas 07 Anomalies

The document discusses improving techniques for independent verification and validation (IV&V) through the analysis of anomalies. It presents an approach using an incremental discretizer and Bayesian classifier to track anomalies in data over time. The approach was tested on flight simulator data and detected off-nominal conditions not present in prior data. The document argues that detecting anomalies could help address issues like those preceding the Challenger and Columbia disasters. It identifies opportunities to apply the techniques for fast-time monitoring in flight simulators and slow-time monitoring of software project data from IV&V to detect any unreported issues. The next steps proposed are to integrate the algorithms with active NASA data sources to assess if anomalies can be detected and repairs proposed.