ABSTRACT
A neural trigger module for fast classification of high-energy physics events, based on the analog neural microprocessor ETANN by Intel, has been realized. It has been employed in the experiment WA92 at CERN. Task of the neural trigger is to send events to be analyzed with high priority on a special data stream, the Fast Stream. Its response time is less than 10 microseconds. The limited precision of the ETANN chip affects the choice of the neural net architecture, as it is discussed in some detail. Results from the experimental run are presented.
