... line.

It would sound like a good idea to use a specialized pattern finder program based on a neural net to identify a cosmic muon event. This however is very difficult because the net would be far too big - it would have to contain all ZEUS detector elements as entry neurons and would thus require a huge number of cosmic muon events to train. But then it would only be able to recognize single cosmic muon events, a detection of a cosmic muon overlapping with a beam-gas or true collision event would be extremely difficult.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.