Anomaly detection with a fruit-fly brain
Could a fruit fly's brain do anomaly detection?
A fruit fly has a tiny brain, yet it notices when a smell is new. It does not store every smell separately; each familiar smell just presses down a few synapses a little more. Industrial defect inspection does the opposite: it stores whole patches of normal images and compares against them. If we switched to the fly's way, memory would shrink by hundreds of times. Would defects still be found?
