Bound on the number of bits per key the function f is allowed (m is the size of the in set, C is the size of the learning function) alpha is the threashold, FP and Fn are false pos and neg rates, and b is the bits per key. For an improvement, this equation must hold. Code it up.
Final pathway: Build GBF and get size in c++, export to python (or use math) -> parameterize classifier -> pre check size -> train -> set t ->determine F_p, F_n, -> check constraint ->export and benchmark
Criteria
Bound on the number of bits per key the function f is allowed (m is the size of the in set, C is the size of the learning function) alpha is the threashold, FP and Fn are false pos and neg rates, and b is the bits per key. For an improvement, this equation must hold. Code it up.
Final pathway: Build GBF and get size in c++, export to python (or use math) -> parameterize classifier -> pre check size -> train -> set t ->determine F_p, F_n, -> check constraint ->export and benchmark