Quick search for rare events through adaptive group sampling

Rare events can potentially occur in many applications and when they model transient opportunities or costly risks should be detected quickly. Due to their sporadic nature, the information-bearing signals associated with rare events often lie in a large set of irrelevant signals and are not easily accessible. This paper provides a sequential search framework that initially takes rough mixed measurements from a group of events. The groups of events that are deemed to be including one or more rare events are retained for further scrutiny and the individual events in such groups are processed sequentially in order to identify a rare event with the shortest delay. Particular focus is placed on Gaussian signals with the aim of detecting signals with rare mean and variance values.