Searching for Anomalies Over Composite Hypotheses Article Swipe
YOU?
·
· 2020
· Open Access
·
· DOI: https://doi.org/10.1109/tsp.2020.2971438
· OA: W2986276296
The problem of detecting anomalies in multiple processes is considered. We\nconsider a composite hypothesis case, in which the measurements drawn when\nobserving a process follow a common distribution with an unknown parameter\n(vector), whose value lies in normal or abnormal parameter spaces, depending on\nits state. The objective is a sequential search strategy that minimizes the\nexpected detection time subject to an error probability constraint. We develop\na deterministic search algorithm with the following desired properties. First,\nwhen no additional side information on the process states is known, the\nproposed algorithm is asymptotically optimal in terms of minimizing the\ndetection delay as the error probability approaches zero. Second, when the\nparameter value under the null hypothesis is known and equal for all normal\nprocesses, the proposed algorithm is asymptotically optimal as well, with\nbetter detection time determined by the true null state. Third, when the\nparameter value under the null hypothesis is unknown, but is known to be equal\nfor all normal processes, the proposed algorithm is consistent in terms of\nachieving error probability that decays to zero with the detection delay.\nFinally, an explicit upper bound on the error probability under the proposed\nalgorithm is established for the finite sample regime. Extensive experiments on\nsynthetic dataset and DARPA intrusion detection dataset are conducted,\ndemonstrating strong performance of the proposed algorithm over existing\nmethods.\n
