Temporal Pattern Recognition
Abstract
A self-organizing network architecture for the learning of recognition codes corresponding to temporal patterns is described. The problem presents itself in many real-world situations. In any non-trivial environment in which a proposed system will function the spectre of temporal information- information coming into the system over a period of time-is evident. In many cases it is not sufficient to process the information independent of its relative time-order. Disciplines as diverse as speech recognition, robotics and data fusion/situation analysis require that temporal aspect of the data be considered. In temporal environments such as these the information lost when using a non-temporal approach can be prohibitive. This approach is formulated to make use of this important temporal information. The network described takes as its input individual incoming events. Sequences of these events (letters, phonemes, or, more abstractly, object sightings in a vision system), received by the system over time are categorized as specific sequences by the temporal system. The Temporal system produces Gaussian classifications that represent the statistics of the temporal data, and the temporal system. The Temporal system produces Gaussian classifications that represent the statistics of the temporal data, and the system uses a noisy environment, giving as output a Gaussian distance from the stored sequence, thus providing an analog measure of closeness of fit to currently known patterns. The system can recognize sequences with missing or extraneous elements, as well as out-of-order sequences. Keywords: Artificial intelligence, Command and control systems, Data fusion, Electro optical, Optoelectronic devices.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 1988
- Accession Number
- ADA200090
Entities
People
- C. E. Priebe
- C. H. Sung