An Environmentally Adaptive, Nonparametric Approach to Some Classification Problems in Underwater Acoustics.
TEXAS UNIV AT AUSTIN APPLIED RESEARCH LABS
Pagination or Media Count:
The subject to this report is the classification of received signals which result from the scattering of underwater acoustic waves. The specific problem area of investigation is a part of the more general domain of remote sensing of an environment. The fundamental influence of environmental conditions and intrinsic properties of the object under study on the correct classification of received waveforms is explicitly incorporated by use of suitably defined environmental state variables. Conventional approaches to this problem employ composite hypothesis tests involving ensemble averages over the various environmental states. It is shown here that alteration of the classification algorithm by introducing available, pertinent environmental state information leads to a reduction i.e., an improvement in the expected loss to this user. This procedure for alteration is called environmental adaptation, and the procedure itself is referred to as an environmentally adaptive approach to classification.
- Acoustic Detection and Detectors
- Non-Radio Communications