def test_weights():
ogdlr_weights = ogdlr_after.weights()
ftrl_weights = ftrl_after.weights()
hash_keys = [mmh(key, seed=SEED) % NDIMS for key in ftrl_after.keys()]
hash_weights = hash_after._get_w(hash_keys)
assert np.allclose(ogdlr_weights, ftrl_weights)
assert np.allclose(hash_weights, ftrl_weights)
def test_weights():
ogdlr_weights = ogdlr_after.weights()
ftrl_weights = ftrl_after.weights()
hash_keys = [mmh(key, seed=SEED) % NDIMS for key in ftrl_after.keys()]
hash_weights = hash_after._get_w(hash_keys)
assert np.allclose(ogdlr_weights, ftrl_weights)
assert np.allclose(hash_weights, ftrl_weights)
def hash_bin_to_bin(self, bin_id, attempt, seed):
key = str(attempt) + "." + str(bin_id)
return mmh(key=key, seed=seed, positive=True) % self.K
def hash_func(self, seed):
return lambda x: mmh(key=x, seed=seed, positive=True) % self.D
if __name__ == '__main__':
s1 = [480, 923, 106]
s2 = [480, 106, 373]
D = 1000
HD = 100
K = 10
print(
"Jaccard",
len([a for a in s1 if a in s2]) /
(len(s1) + len(s2) - len([a for a in s1 if a in s2])))
vs = []
for i in range(0, 1000):
#DMH = Densified_MinHash(K, HD, seed=i)
DMH = Densified_MinHash(K,
HD,
seed=mmh(i, 30, positive=True),
num_seed=mmh(i, 20, positive=True),
hashFull=True)
#x1 = DMH.get_hashed(s1)
#x2 = DMH.get_hashed(s2)
#xs1 = DMH.convert_to_bit_array(x1)
#xs2 = DMH.convert_to_bit_array(x2)
#xs1 = DMH.get_hashed_faster(s1)
#xs2 = DMH.get_hashed_faster(s2)
xs1 = DMH.get_hashed_4universal(s1)
xs2 = DMH.get_hashed_4universal(s2)
vs.append(np.dot(xs1, xs2) / K)
print(np.mean(vs), np.std(vs))
def ft_mmh(text): global m hash_value = mmh(text, positive = True) return (hash_value % m)