def fit(self, X):
hashes = []
for k in xrange(self._hash_counts):
nearpy_rbp = nearpy.hashes.RandomBinaryProjections('rbp_%d' % k, self._n_bits)
hashes.append(nearpy_rbp)
self._nearpy_engine = nearpy.Engine(X.shape[1], lshashes=hashes, distance=nearpy.distances.CosineDistance())
for i, x in enumerate(X):
self._nearpy_engine.store_vector(x.tolist(), i)
def fit(self, X):
hashes = []
# TODO: doesn't seem like the NearPy code is using the metric??
for k in xrange(self._hash_counts):
nearpy_rbp = nearpy.hashes.RandomBinaryProjections('rbp_%d' % k, self._n_bits)
hashes.append(nearpy_rbp)
self._nearpy_engine = nearpy.Engine(X.shape[1], lshashes=hashes)
for i, x in enumerate(X):
self._nearpy_engine.store_vector(x.tolist(), i)
def fit(self, X):
hashes = []
for k in xrange(self._hash_counts):
nearpy_rbp = nearpy.hashes.RandomBinaryProjections(
'rbp_%d' % k, self._n_bits)
hashes.append(nearpy_rbp)
self._nearpy_engine = nearpy.Engine(X.shape[1], lshashes=hashes)
for i, x in enumerate(X):
self._nearpy_engine.store_vector(x.tolist(), i)
def fit(self, X):
import nearpy, nearpy.hashes, nearpy.distances
hashes = []
# TODO: doesn't seem like the NearPy code is using the metric??
for k in xrange(self._hash_counts):
nearpy_rbp = nearpy.hashes.RandomBinaryProjections(
'rbp_%d' % k, self._n_bits)
hashes.append(nearpy_rbp)
self._nearpy_engine = nearpy.Engine(X.shape[1], lshashes=hashes)
for i, x in enumerate(X):
self._nearpy_engine.store_vector(x.tolist(), i)
def handle(self, *args, **options):
import nearpy, nearpy.hashes, nearpy.distances
resample_dim_pool = [[4,4,4],[6,6,6],[8,8,8],[10,10,10],[12,12,12],[14,14,14],[16,16,16]]
subjects = 940
n_bits_pool = [2, 4, 6, 8, 10]
hash_counts_pool = [2, 5, 10, 15, 20]
metric_pool = ["euclidean","cosine"]
z_score_pool = ["yes", "no"]
for resample_dim in resample_dim_pool:
features, dict_feat = createFeatures(subjects, resample_dim)
scores = get_neurovault_scores(subjects, dict_feat)
for n_bits in n_bits_pool:
for hash_counts in hash_counts_pool:
for metric in metric_pool:
#for z_score in z_score_pool:
if metric == "euclidean":
distance = nearpy.distances.EuclideanDistance()
else:
distance = nearpy.distances.EuclideanDistance()
# fit
hashes = []
for k in xrange(hash_counts):
nearpy_rbp = nearpy.hashes.RandomBinaryProjections('rbp_%d' % k, n_bits)
hashes.append(nearpy_rbp)
nearpy_engine = nearpy.Engine(features.shape[1], lshashes=hashes, distance=distance)
for i, x in enumerate(features):
nearpy_engine.store_vector(x.tolist(), dict_feat[i])
#query
for i in range(features.shape[0]):
results = nearpy_engine.neighbours(features[i])
features, dict_feat = np.load('/code/neurovault/apps/statmaps/tests/features_940_16_16_16.npy').T, \
pickle.load(open('/code/neurovault/apps/statmaps/tests/dict_feat.p', "rb"))
import timeit
## NEARPY TEST
n_bits = 5
hash_counts = 20
metric = "euclidean"
name = 'NearPy(n_bits=%d, hash_counts=%d)' % (n_bits, hash_counts)
# fiting
import nearpy, nearpy.hashes, nearpy.distances
hashes = []
# doesn't seem like the NearPy code is using the metric??
for k in xrange(hash_counts):
nearpy_rbp = nearpy.hashes.RandomBinaryProjections('rbp_%d' % k, n_bits)
hashes.append(nearpy_rbp)
filter_N = nearpy.filters.NearestFilter(100)
nearpy_engine = nearpy.Engine(features.shape[1], distance= nearpy.distances.EuclideanDistance(),
lshashes=hashes,vector_filters=[filter_N])
#indexing
t = Timer()
with t:
for i, x in enumerate(features):
nearpy_engine.store_vector(x.tolist(), dict_feat[i])
# querying
for i in range(features.shape[0]):
t = Timer()
with t:
results = nearpy_engine.neighbours(features[i])