Revision history [back]

all 3 algorithms are using a simple (1) Nearest Neighbour classification, like this:

for each image in trainset:
    if distance between train & test img is the smallest so far:
          keep best id & distance
return best id & distance