class Picker(object):
# 结合 K-Means 里的方法来挑选隐藏层中心点
kmeans = KMeans()
samples = [] # <Sample Object>
# Random Picking
def shuffing(self, k):
maker = self.kmeans.center_maker
maker.samples = self.samples
return maker.shuffing(k)
# K-Means++ 方法挑选后做聚类更新中心点（预设欧式距离）
def clustering(self, k):
# 先把 Sample Object 的 features 都取出来做聚类（Clustering）
for sample in self.samples:
self.kmeans.add_sample(sample.features)
kmeans = self.kmeans
kmeans.center_choice = Choice.Plus
kmeans.make_centers(k)
kmeans.max_iteration = 100
kmeans.convergence = 0.001
kmeans.setup()
kmeans.run()
# 将聚好且修正学习过好的中心点重新做回去 <Sample Object>，回传回去给 RBFNN 使用
clustered_samples = [] # <Sample Object>
for group in kmeans.groups:
sample = Sample(group.center) # 这里的 Sample Object 不需要设定 targets
clustered_samples.append(sample)
return clustered_samples
class Test(unittest.TestCase):
def setUp(self):
self.km = KMeans(iterations=10,useLUT=True)
def testIsRGBA(self):
fi=open("Mona_Lisa.jpg","r")
isRGBA = self.km.testRGBA(Image.open(fi))
fi.close()
assert not isRGBA
def testProcessFile(self):
fi=open("rusty door.jpg","r")
fo=open("result.jpg","w")
self.km.process(fi,fo,centroids=10)
fo.close()
fi.close()
totalnumberofpixels = self.km.im.size[0]*self.km.im.size[1]
assert len(self.km.pixels) == totalnumberofpixels
for x in self.km.pixels:
#print x
r,g,b,centroid = x
assert r in range(0,256)
assert g in range(0,256)
assert b in range(0,256)
assert centroid in range(0,self.km.numbercentroids)
totpixels=0
assert len(self.km.clusters.keys())==self.km.numbercentroids
for x in self.km.clusters.keys():
assert len(self.km.clusters[x])>0
totpixels+= len(self.km.clusters[x])
print "Cluster %d has %d elements" % (x,len(self.km.clusters[x]))
assert totpixels == totalnumberofpixels
for x in range(0,len(self.km.means)):
r,g,b = self.km.means[x]
assert r in range(0,256)
assert g in range(0,256)
assert b in range(0,256)
print "Cluster %d is centered at (%d,%d,%d)" % (x,r,g,b)
def construct_kmeans_obj(collection, level):
"""
Grabs the kmeans data from the db
Re-creates the KMeans object
Data:
{
'class_name': 'doesn't matter,
'k': 3,
'labels': ['achiever', 'collector', 'killer']
'centroids': [ [] [] [] ],
'clusters': [ [] [] [] ],
}
"""
k = None
k_data = collection.find_one({'level': level})
if k_data is not None:
k = KMeans(k=k_data['k'], class_name=k_data['class_name'])
k.centroids = k_data['centroids']
k.clusters = k_data['clusters']
k.labels = k_data['labels']
return k
def initialise(self, init_FG='random', init_S='random'):
''' Initialise F, S, G, tau, and lambdaFk, lambdaGl (if ARD). '''
assert init_FG in OPTIONS_INIT_FG, "Unknown initialisation option for F and G: %s. Should be in %s." % (
init_FG, OPTIONS_INIT_FG)
assert init_S in OPTIONS_INIT_S, "Unknown initialisation option for S: %s. Should be in %s." % (
init_S, OPTIONS_INIT_S)
# Initialise lambdaFk, lambdaGl, and compute expectations
if self.ARD:
self.alphaFk_s, self.betaFk_s = numpy.zeros(self.K), numpy.zeros(
self.K)
self.alphaGl_s, self.betaGl_s = numpy.zeros(self.L), numpy.zeros(
self.L)
self.exp_lambdaFk, self.exp_loglambdaFk = numpy.zeros(
self.K), numpy.zeros(self.K)
self.exp_lambdaGl, self.exp_loglambdaGl = numpy.zeros(
self.L), numpy.zeros(self.L)
for k in range(self.K):
self.alphaFk_s[k] = self.alpha0
self.betaFk_s[k] = self.beta0
self.update_exp_lambdaFk(k)
for l in range(self.L):
self.alphaGl_s[l] = self.alpha0
self.betaGl_s[l] = self.beta0
self.update_exp_lambdaGl(l)
# Initialise parameters F, G
self.mu_F, self.tau_F = numpy.zeros((self.I, self.K)), numpy.zeros(
(self.I, self.K))
self.mu_G, self.tau_G = numpy.zeros((self.J, self.L)), numpy.zeros(
(self.J, self.L))
self.mu_S, self.tau_S = numpy.zeros((self.K, self.L)), numpy.zeros(
(self.K, self.L))
if init_FG == 'kmeans':
print "Initialising F using KMeans."
kmeans_F = KMeans(self.R, self.M, self.K)
kmeans_F.initialise()
kmeans_F.cluster()
self.mu_F = kmeans_F.clustering_results
for i, k in itertools.product(range(self.I), range(self.K)):
self.tau_F[i, k] = 1.
print "Initialising G using KMeans."
kmeans_G = KMeans(self.R.T, self.M.T, self.L)
kmeans_G.initialise()
kmeans_G.cluster()
self.mu_G = kmeans_G.clustering_results
for j, l in itertools.product(range(self.J), range(self.L)):
self.tau_G[j, l] = 1.
else:
# 'random' or 'exp'
for i, k in itertools.product(range(self.I), range(self.K)):
self.tau_F[i, k] = 1.
hyperparam = self.exp_lambdaFk[
k] if self.ARD else self.lambdaF[i, k]
self.mu_F[i, k] = exponential_draw(
hyperparam) if init_FG == 'random' else 1.0 / hyperparam
for j, l in itertools.product(range(self.J), range(self.L)):
self.tau_G[j, l] = 1.
hyperparam = self.exp_lambdaGl[
l] if self.ARD else self.lambdaG[j, l]
self.mu_G[j, l] = exponential_draw(
hyperparam) if init_FG == 'random' else 1.0 / hyperparam
# Initialise parameters S
for k, l in itertools.product(range(self.K), range(self.L)):
self.tau_S[k, l] = 1.
hyperparam = self.lambdaS[k, l]
self.mu_S[k, l] = exponential_draw(
hyperparam) if init_S == 'random' else 1.0 / hyperparam
# Compute expectations and variances F, G, S
self.exp_F, self.var_F = numpy.zeros((self.I, self.K)), numpy.zeros(
(self.I, self.K))
self.exp_G, self.var_G = numpy.zeros((self.J, self.L)), numpy.zeros(
(self.J, self.L))
self.exp_S, self.var_S = numpy.zeros((self.K, self.L)), numpy.zeros(
(self.K, self.L))
for k in range(self.K):
self.update_exp_F(k)
for l in range(self.L):
self.update_exp_G(l)
for k, l in itertools.product(range(self.K), range(self.L)):
self.update_exp_S(k, l)
# Initialise tau and compute expectation
self.update_tau()
self.update_exp_tau()
def initialise(self,init_S='random',init_FG='random',tauFSG={}):
self.tauF = tauFSG['tauF'] if 'tauF' in tauFSG else numpy.ones((self.I,self.K))
self.tauS = tauFSG['tauS'] if 'tauS' in tauFSG else numpy.ones((self.K,self.L))
self.tauG = tauFSG['tauG'] if 'tauG' in tauFSG else numpy.ones((self.J,self.L))
assert init_S in ['exp','random'], "Unrecognised init option for S: %s." % init_S
self.muS = 1./self.lambdaS
if init_S == 'random':
for k,l in itertools.product(xrange(0,self.K),xrange(0,self.L)):
self.muS[k,l] = exponential_draw(self.lambdaS[k,l])
assert init_FG in ['exp','random','kmeans'], "Unrecognised init option for F,G: %s." % init_FG
self.muF, self.muG = 1./self.lambdaF, 1./self.lambdaG
if init_FG == 'random':
for i,k in itertools.product(xrange(0,self.I),xrange(0,self.K)):
self.muF[i,k] = exponential_draw(self.lambdaF[i,k])
for j,l in itertools.product(xrange(0,self.J),xrange(0,self.L)):
self.muG[j,l] = exponential_draw(self.lambdaG[j,l])
elif init_FG == 'kmeans':
print "Initialising F using KMeans."
kmeans_F = KMeans(self.R,self.M,self.K)
kmeans_F.initialise()
kmeans_F.cluster()
self.muF = kmeans_F.clustering_results #+ 0.2
print "Initialising G using KMeans."
kmeans_G = KMeans(self.R.T,self.M.T,self.L)
kmeans_G.initialise()
kmeans_G.cluster()
self.muG = kmeans_G.clustering_results #+ 0.2
# Initialise the expectations and variances
self.expF, self.varF = numpy.zeros((self.I,self.K)), numpy.zeros((self.I,self.K))
self.expS, self.varS = numpy.zeros((self.K,self.L)), numpy.zeros((self.K,self.L))
self.expG, self.varG = numpy.zeros((self.J,self.L)), numpy.zeros((self.J,self.L))
for k in range(0,self.K):
self.update_exp_F(k)
for k,l in itertools.product(xrange(0,self.K),xrange(0,self.L)):
self.update_exp_S(k,l)
for l in range(0,self.L):
self.update_exp_G(l)
# Initialise tau using the updates
self.update_tau()
#self.alpha_s, self.beta_s = self.alpha, self.beta
self.update_exp_tau()
def initialise(self, init_S='random', init_FG='random'):
assert init_S in [
'random', 'exp'
], "Unknown initialisation option for S: %s. Should be 'random' or 'exp'." % init_S
assert init_FG in [
'random', 'exp', 'kmeans'
], "Unknown initialisation option for S: %s. Should be 'random', 'exp', or 'kmeans." % init_FG
self.S = 1. / self.lambdaS
if init_S == 'random':
for k, l in itertools.product(xrange(0, self.K), xrange(0,
self.L)):
self.S[k, l] = exponential_draw(self.lambdaS[k, l])
self.F, self.G = 1. / self.lambdaF, 1. / self.lambdaG
if init_FG == 'random':
for i, k in itertools.product(xrange(0, self.I), xrange(0,
self.K)):
self.F[i, k] = exponential_draw(self.lambdaF[i, k])
for j, l in itertools.product(xrange(0, self.J), xrange(0,
self.L)):
self.G[j, l] = exponential_draw(self.lambdaG[j, l])
elif init_FG == 'kmeans':
print "Initialising F using KMeans."
kmeans_F = KMeans(self.R, self.M, self.K)
kmeans_F.initialise()
kmeans_F.cluster()
self.F = kmeans_F.clustering_results + 0.2
print "Initialising G using KMeans."
kmeans_G = KMeans(self.R.T, self.M.T, self.L)
kmeans_G.initialise()
kmeans_G.cluster()
self.G = kmeans_G.clustering_results + 0.2
self.tau = self.alpha_s() / self.beta_s()
def initialise(self, init_FG='random', init_S='random'):
''' Initialise F, S, G, tau, and lambdaFk, lambdaGl (if ARD). '''
assert init_FG in OPTIONS_INIT_FG, "Unknown initialisation option for F and G: %s. Should be in %s." % (
init_FG, OPTIONS_INIT_FG)
assert init_S in OPTIONS_INIT_S, "Unknown initialisation option for S: %s. Should be in %s." % (
init_S, OPTIONS_INIT_S)
self.F = numpy.zeros((self.I, self.K))
self.S = numpy.zeros((self.K, self.L))
self.G = numpy.zeros((self.J, self.L))
self.lambdaFk = numpy.zeros(self.K)
self.lambdaGl = numpy.zeros(self.L)
# Initialise lambdaFk, lambdaGl
if self.ARD:
for k in range(self.K):
self.lambdaFk[k] = self.alpha0 / self.beta0
for l in range(self.L):
self.lambdaGl[l] = self.alpha0 / self.beta0
# Initialise F, G
if init_FG == 'kmeans':
print "Initialising F using KMeans."
kmeans_F = KMeans(self.R, self.M, self.K)
kmeans_F.initialise()
kmeans_F.cluster()
self.F = kmeans_F.clustering_results + 0.2
print "Initialising G using KMeans."
kmeans_G = KMeans(self.R.T, self.M.T, self.L)
kmeans_G.initialise()
kmeans_G.cluster()
self.G = kmeans_G.clustering_results + 0.2
else:
# 'random' or 'exp'
for i, k in itertools.product(range(self.I), range(self.K)):
hyperparam = self.lambdaFk[k] if self.ARD else self.lambdaF[i,
k]
self.F[i, k] = exponential_draw(
hyperparam) if init_FG == 'random' else 1.0 / hyperparam
for j, l in itertools.product(range(self.J), range(self.L)):
hyperparam = self.lambdaGl[l] if self.ARD else self.lambdaG[j,
l]
self.G[j, l] = exponential_draw(
hyperparam) if init_FG == 'random' else 1.0 / hyperparam
# Initialise S
for k, l in itertools.product(range(self.K), range(self.L)):
hyperparam = self.lambdaS[k, l]
self.S[k, l] = exponential_draw(
hyperparam) if init_S == 'random' else 1.0 / hyperparam
# Initialise tau
self.tau = gamma_draw(self.alpha_s(), self.beta_s())
def initialise(self, init_FG='random', init_S='random', expo_prior=1.):
''' Initialise F, S and G. '''
assert init_FG in OPTIONS_INIT_FG, "Unrecognised init option for F,G: %s. Should be one in %s." % (
init_FG, OPTIONS_INIT_FG)
assert init_S in OPTIONS_INIT_S, "Unrecognised init option for S: %s. Should be one in %s." % (
init_S, OPTIONS_INIT_S)
if init_S == 'ones':
self.S = numpy.ones((self.K, self.L))
elif init_S == 'random':
self.S = numpy.random.rand(self.K, self.L)
elif init_S == 'exponential':
self.S = numpy.empty((self.K, self.L))
for k, l in itertools.product(xrange(0, self.K), xrange(0,
self.L)):
self.S[k, l] = exponential_draw(expo_prior)
if init_FG == 'ones':
self.F = numpy.ones((self.I, self.K))
self.G = numpy.ones((self.J, self.L))
elif init_FG == 'random':
self.F = numpy.random.rand(self.I, self.K)
self.G = numpy.random.rand(self.J, self.L)
elif init_FG == 'exponential':
self.F = numpy.empty((self.I, self.K))
self.G = numpy.empty((self.J, self.L))
for i, k in itertools.product(xrange(0, self.I), xrange(0,
self.K)):
self.F[i, k] = exponential_draw(expo_prior)
for j, l in itertools.product(xrange(0, self.J), xrange(0,
self.L)):
self.G[j, l] = exponential_draw(expo_prior)
elif init_FG == 'kmeans':
print "Initialising F using KMeans."
kmeans_F = KMeans(self.R, self.M, self.K)
kmeans_F.initialise()
kmeans_F.cluster()
self.F = kmeans_F.clustering_results + 0.2
print "Initialising G using KMeans."
kmeans_G = KMeans(self.R.T, self.M.T, self.L)
kmeans_G.initialise()
kmeans_G.cluster()
self.G = kmeans_G.clustering_results + 0.2
def setUp(self):
self.km = KMeans(iterations=10,useLUT=True)
from typing import List
from kmeans.data_point import DataPoint
from kmeans.kmeans import KMeans
if __name__ == "__main__":
point1: DataPoint = DataPoint([2.0, 1.0, 1.0])
point2: DataPoint = DataPoint([2.0, 2.0, 5.0])
point3: DataPoint = DataPoint([3.0, 1.5, 2.5])
kmeans_test: KMeans[DataPoint] = KMeans(2, [point1, point2, point3])
test_clusters: List[KMeans.Cluster] = kmeans_test.run()
for index, cluster in enumerate(test_clusters):
print(f"Cluster {index}: {cluster.points}")
parser.add_argument('-c', help='Max iterations before convergence', type=int, default=300)
parser.add_argument('-t', help="Number of trials", type=int, default=1)
parser.add_argument('-k', help='Value of \'k\' (centroids)', required=True, type=int, default=2)
parser.add_argument('-n', help='data size', required=True, type=int, default=1000)
# Get/parse arguments
args = parser.parse_args()
# Number of data points
no_points = args.n
with open('./trials/time.{0}.{1}.{2}.csv'.format(args.n, args.k, args.t), 'wb') as f:
f.write('data size,k,trial,kmeans time (s)\n')
for trial in xrange(1, args.t + 1):
# KMeans Object
k = KMeans(class_name="test", k=args.k, log=args.log, max_iterations=args.c)
# Generate Data
data = [
[
random.uniform(0.0, 1.0), # time
random.uniform(0.0, 1.5), # coins, can be up to 1.5 the ammount of coins
random.uniform(0.0, 1.0) # kills
] for i in range(no_points)
]
# Inster data (this forces KMeans calculation)
start = time.clock()
k.put(data)
f.write('{0},{1},{2},{3}\n'.format(args.n, args.k, trial, (time.clock() - start) ))
from kmeans.data import Data
from kmeans.settings import Settings
from kmeans.kmeans import KMeans
if __name__ == '__main__':
settings = Settings()
settings.read_parameters()
settings.print_parameters()
data = Data(directory='../data/', data='data.txt')
km = KMeans(data, settings)
km.run()
