def load(self, filename):
with open(filename, "rb") as pkl_file:
this_d = pickle.load(pkl_file)
self.__dict__.update(this_d)
if "indexName" in this_d:
I = KGraph()
I.load(self.indexName)
self.Index = I
def load(self, filename):
with open(filename, 'rb') as pkl_file:
this_d = pickle.load(pkl_file)
self.__dict__.update(this_d)
if 'indexName' in this_d:
I = KGraph()
I.load(self.indexName)
self.Index = I
def __init__(self):
self.Index = KGraph()
self.iterations = 20
self.L = 64
self.K = 8
self.S = 30
self.controls = 100
self.delta = 0.005
self.recall = 0.95
self.prune = 2
class KnnIndex:
def __init__(self):
self.Index = KGraph()
self.iterations = 20
self.L = 64
self.K = 8
self.S = 30
self.controls = 100
self.delta = 0.005
self.recall = 0.95
self.prune = 2
def set_dataset_key_map(self, map_vector):
assert len(self.dataset) == len(map_vector)
self.key_map = map_vector
def train(self, dataset):
self.dataset = dataset
self.dtype = dataset.dtype
assert self.dtype in [np.uint8, np.float]
self.Index.build(
self.dataset, self.iterations, self.L, self.K, self.S, self.controls, self.delta, self.recall, self.prune
)
def save(self, filename):
indexName = filename.split(".")[0] + ".kgi"
self.indexName = indexName
self.Index.save(indexName)
d = dict(self.__dict__)
del d["Index"]
with open(filename, "wb") as fp:
pickle.dump(d, fp)
return filename
def load(self, filename):
with open(filename, "rb") as pkl_file:
this_d = pickle.load(pkl_file)
self.__dict__.update(this_d)
if "indexName" in this_d:
I = KGraph()
I.load(self.indexName)
self.Index = I
def query(self, queryset, K=8, probe=100, mapped=False):
assert self.dtype == queryset.dtype
ans = self.Index.search(self.dataset, queryset, K=K, P=probe)
if mapped:
return np.array([[self.key_map[i] for i in line] for line in ans])
return np.array(ans)
class KnnIndex:
def __init__(self):
self.Index = KGraph()
self.iterations = 20
self.L = 64
self.K = 8
self.S = 30
self.controls = 100
self.delta = 0.005
self.recall = 0.95
self.prune = 2
def set_dataset_key_map(self, map_vector):
assert (len(self.dataset) == len(map_vector))
self.key_map = map_vector
def train(self, dataset):
self.dataset = dataset
self.dtype = dataset.dtype
assert (self.dtype in [np.uint8, np.float])
self.Index.build(self.dataset, self.iterations, self.L, self.K, self.S,
self.controls, self.delta, self.recall, self.prune)
def save(self, filename):
indexName = filename.split('.')[0] + '.kgi'
self.indexName = indexName
self.Index.save(indexName)
d = dict(self.__dict__)
del d['Index']
with open(filename, 'wb') as fp:
pickle.dump(d, fp)
return filename
def load(self, filename):
with open(filename, 'rb') as pkl_file:
this_d = pickle.load(pkl_file)
self.__dict__.update(this_d)
if 'indexName' in this_d:
I = KGraph()
I.load(self.indexName)
self.Index = I
def query(self, queryset, K=8, probe=100, mapped=False):
assert (self.dtype == queryset.dtype)
ans = self.Index.search(self.dataset, queryset, K=K, P=probe)
if mapped:
return np.array([[self.key_map[i] for i in line] for line in ans])
return np.array(ans)
def __init__(self):
self.Index = KGraph()
self.iterations = 20
self.L = 64
self.K = 8
self.S = 30
self.controls = 100
self.delta = 0.005
self.recall = 0.95
self.prune = 2
def build_index(self, X):
index = KGraph()
index.build(X)
return index
def build_index(self, X):
index = KGraph()
index.build(X)
return index
