If I understand you correctly, RandomBinaryProjection is a hashing of the cosine distance. In general, there are different hashing methods for different types of distances. Does the NearPy library not provide hashing methods for Euclidean distance or Hamming distance? Why run with different types of distances during the filtering phase?
If I understand you correctly,
RandomBinaryProjectionis a hashing of the cosine distance. In general, there are different hashing methods for different types of distances. Does the NearPy library not provide hashing methods for Euclidean distance or Hamming distance? Why run with different types of distances during the filtering phase?