Classifying and validating intermittent EEG patterns with syntactic methods

摘要

For diagnostic purposes the EEG is recorded as a multichannel signal. For classifying and validating intermittent EEG patterns, the temporal and spatial relations between the constituent basic patterns are important. When attempting syntactic pattern recognition from these patterns, the absence of a natural string representation causes problems. They were solved by using a type of attributed grammar and putting all spatial and temporal information into the attributes. Three of these grammars are used in a production system together with a scheduling algorithm. The first results from this system are given.