New Algorithm Mimics Fruit Fly’s Long‑Term Smell Memory

Researchers at the Okinawa Institute of Science and Technology have released Spi‑Fly, an algorithm that emulates the olfactory system of the fruit fly, which contains about 140,000 neurons. The model can identify a wide range of odors in a fraction of a second and retain the memory of each scent for an extended period. Current electronic noses are costly, detect only a narrow set of chemicals, and forget odors as soon as new ones are learned. Spi‑Fly’s design promises to overcome these limitations by providing rapid, persistent odor recognition without the need for large, expensive sensor arrays. The algorithm was published in the journal Neuromorphic Computing and Engineering.