self_organizing_map.py

"""
https://en.wikipedia.org/wiki/Self-organizing_map
"""

importmath


classSelfOrganizingMap:
defget_winner(self, weights: list[list[float]], sample: list[int]) ->int:
"""
 Compute the winning vector by Euclidean distance

 >>> SelfOrganizingMap().get_winner([[1, 2, 3], [4, 5, 6]], [1, 2, 3])
 1
 """
d0=0.0
d1=0.0
foriinrange(len(sample)):
d0+=math.pow((sample[i] -weights[0][i]), 2)
d1+=math.pow((sample[i] -weights[1][i]), 2)
return0ifd0>d1else1
return0

defupdate(
self, weights: list[list[int|float]], sample: list[int], j: int, alpha: float
 ) ->list[list[int|float]]:
"""
 Update the winning vector.

 >>> SelfOrganizingMap().update([[1, 2, 3], [4, 5, 6]], [1, 2, 3], 1, 0.1)
 [[1, 2, 3], [3.7, 4.7, 6]]
 """
foriinrange(len(weights)):
weights[j][i] +=alpha* (sample[i] -weights[j][i])
returnweights


# Driver code
defmain() ->None:
# Training Examples ( m, n )
training_samples= [[1, 1, 0, 0], [0, 0, 0, 1], [1, 0, 0, 0], [0, 0, 1, 1]]

# weight initialization ( n, C )
weights= [[0.2, 0.6, 0.5, 0.9], [0.8, 0.4, 0.7, 0.3]]

# training
self_organizing_map=SelfOrganizingMap()
epochs=3
alpha=0.5

for_inrange(epochs):
forjinrange(len(training_samples)):
# training sample
sample=training_samples[j]

# Compute the winning vector
winner=self_organizing_map.get_winner(weights, sample)

# Update the winning vector
weights=self_organizing_map.update(weights, sample, winner, alpha)

# classify test sample
sample= [0, 0, 0, 1]
winner=self_organizing_map.get_winner(weights, sample)

# results
print(f"Clusters that the test sample belongs to : {winner}")
print(f"Weights that have been trained : {weights}")


# running the main() function
if__name__=="__main__":
main()