There’s a great article on PLOS ONE about using pigeons for cancer detection.
Pathologists and radiologists spend years acquiring and refining their medically essential visual skills, so it is of considerable interest to understand how this process actually unfolds and what image features and properties are critical for accurate diagnostic performance. Key insights into human behavioral tasks can often be obtained by using appropriate animal models.
The short version is that scientist successfully trained pigeons to detect cancer through visual inspection of medical imagery. But the kicker for them was not pigeon detection, but rather the pooled results of a flock of pigeons are extremely accurate. This shouldn’t be surprising. It’s the wisdom of crowds applied to pigeons.
It’s also how random forest works. Random trees take a random subset of parameters and creates a decision tree to classify an outcome. A random forest is a “forest” of these random trees. The outcome if whatever receives the most votes. And it’s a huge step forward in data science and classification.
Having a pigeon do it is neat, and it suggests flocks of small neural networks could fill the void.
Image by Andy F / Wikimedia.