def random2HemiPermutations(mat):
"""Return a matrix with elements of a triangle permuted without saving the degrees.
"""
s = mat.shape
N = s[0]
N2 = N / 2
L = N * (N - 2) / 8
nruter = array(mat)
Sind = triSup(
arange(N2 * N2).reshape((N2, N2)) +
arange(0, N2 * N2, N2).reshape((N2, 1)))
Mind = (arange(N2 * N2).reshape(
(N2, N2)) + arange(N2, N2 * N2 + 1, N2).reshape(
(N2, 1))).reshape(N2 * N2)
Iind = triSup(ones((N, N)), ind=1)[-L:]
Salea = permutation(Sind)
Malea = permutation(Mind)
Ialea = permutation(Iind)
nruter[unravel_index(Sind, s)] = nruter[unravel_index(Salea, s)]
nruter[unravel_index(Mind, s)] = nruter[unravel_index(Malea, s)]
nruter[unravel_index(Iind, s)] = nruter[unravel_index(Ialea, s)]
for i in range(s[0]):
nruter[i:, i] = nruter[i, i:]
return nruter
def initialRand(self, dens, normX=None):
'''...'''
self.x = zeros(self.N)
self.A = zeros(self.N)
self.x[permutation(self.N)[range(int(dens * self.N))]] = 1.
self.A[permutation(self.N)[range(int(dens * self.N))]] = 1.
if normX != None:
self.x *= normX / norm(self.x)
def createSimilarAR(data):
"""
creates an AR-process that is similar to a given data set.
data must be given in n x d-format
"""
# step 1: get "average" fit matrix
l_A = []
for rep in arange(100):
idx = randint(0, data.shape[0] - 1, data.shape[0] - 1)
idat = data[idx, :]
odat = data[idx + 1, :]
l_A.append(lstsq(idat, odat)[0])
sysmat = meanMat(l_A).T
# idea: get "dynamic noise" from input data as difference of
# expected vs. predicted data:
# eta_i = (sysmat*(data[:,i-1]).T - data[:,i])
# however, in order to destroy any possible correlations in the
# input noise (they would also occur in the output), the
# noise per section has to be permuted.
prediction = dot(sysmat, data[:-1, :].T)
dynNoise = data[1:, :].T - prediction
res = [
zeros((dynNoise.shape[0], 1)),
]
for nidx in permutation(dynNoise.shape[1]):
res.append(dot(sysmat, res[-1]) + dynNoise[:, nidx][:, newaxis])
return hstack(res).T
def prepVStimDict(self):
res=int(self.params['res'])
if mod(res, 2)==0:
box_order=arange(res**2/2+1, res**2/2+1+res)
else:
box_order=arange(res**2/2.-res/2.+1,res**2/2.-res/2.+1+res)
self.params['box_order']=box_order[permutation(res).astype(int)]
self.vStimDict={'pre':self.params['pre']/1000.,
'dur':self.params['dur']/1000.,
'post':0.1,
'bg':self.params['bg']/100.,
'fg':self.params['fg']/100.,
'reps':self.params['reps'],
'squarenum':self.params['box_order'],
'res':self.params['res'],
'radius':self.params['box_sz']/2,
'sz':self.params['sz'],
'type':'squareboard',
'calProtName':'fastSBLineCal',
'trigger':self.params['trigger']}
self.calMod.calibrateBG(self.vStimDict,self.params,self.infoDict['dirVar'].get())
self.calMod.calibrateFG(self.vStimDict,self.params,self.infoDict['dirVar'].get())
self.notifyStimMachine()
self.RFLineDur=zeros((self.params['reps'],self.params['res'],))
self.RFLinePost=zeros((self.params['reps'],self.params['res'],))
def test_column_set_get():
m = 256
n = 128
data = np.random.randn(m, n)
cm_data = cm.CUDAMatrix(cm.reformat(data))
indices = permutation(n)
cm_indices = cm.CUDAMatrix(cm.reformat(indices.reshape(1, -1)))
start = 0
end = 10
cm_columns = cm_data.get_column_vectors(cm_indices, start, end)
get_error = np.sum((cm_columns.asarray() - data[:, indices[start:end]])**2)
data_set = np.random.randn(m, end - start)
cm_columns.free_device_memory()
cm_columns = cm.CUDAMatrix(cm.reformat(data_set))
cm_data.set_column_vectors(cm_indices, start, end, cm_columns)
data[:, indices[start:end]] = data_set
set_error = np.sum((cm_data.asarray() - data)**2)
print "Get Error = ", get_error
print "Set Error = ", set_error
assert get_error < 10**-2 or set_error < 10**-2, \
"Error in CUDAMatrix.get_column_vectors exceeded threshold"
def permute(lst):
ans = []
from pylab import permutation
p = permutation(len(lst))
for i in range(len(lst)):
ans.append(lst[p[i]])
return ans
def createSimilarAR(data):
"""
creates an AR-process that is similar to a given data set.
data must be given in n x d-format
"""
# step 1: get "average" fit matrix
l_A = []
for rep in arange(100):
idx = randint(0,data.shape[0]-1,data.shape[0]-1)
idat = data[idx,:]
odat = data[idx+1,:]
l_A.append(lstsq(idat,odat)[0])
sysmat = meanMat(l_A).T
# idea: get "dynamic noise" from input data as difference of
# expected vs. predicted data:
# eta_i = (sysmat*(data[:,i-1]).T - data[:,i])
# however, in order to destroy any possible correlations in the
# input noise (they would also occur in the output), the
# noise per section has to be permuted.
prediction = dot(sysmat,data[:-1,:].T)
dynNoise = data[1:,:].T - prediction
res = [zeros((dynNoise.shape[0],1)), ]
for nidx in permutation(dynNoise.shape[1]):
res.append( dot(sysmat,res[-1]) + dynNoise[:,nidx][:,newaxis] )
return hstack(res).T
def main():
XC = loadtxt('iris.data', delimiter=',', dtype=float, converters={4: cnvt})
ind = arange(150) # indices into the dataset
ind = permutation(ind) # random permutation
L = ind[0:90] # learning set indices
T = ind[90:] # test set indices
# Learning Set
X = transpose(XC[L, 0:4])
nnc = NNb(X, XC[L, -1])
# Classification of Test Set
c = zeros(len(T))
for i in arange(len(T)):
print sys.argv[1]
print int(sys.argv[1])
c[i] = nnc.classify(XC[T[i], 0:4], int(sys.argv[1]))
# Confusion Matrix
CM = zeros((3, 3))
for i in range(3):
for j in range(3):
CM[i, j] = sum(logical_and(XC[T, 4] == (i + 1), c == (j + 1)))
print(CM)
# Plot Test Set
figure(1)
color = array([[1, 0, 0], [0, 1, 0], [0, 0, 1]])
for i in range(4):
for j in range(4):
subplot(4, 4, 4 * i + j + 1)
if i == j:
continue
print color[XC[T, 4].astype(int) - 1]
print[1, 1, 1] * len(T)
print color[c.astype(int) - 1]
scatter(XC[T, i],
XC[T, j],
s=100,
marker='s',
edgecolor=color[XC[T, 4].astype(int) - 1],
facecolor=[1, 1, 1] * len(T))
scatter(XC[T, i],
XC[T, j],
s=30,
marker='+',
edgecolor=color[c.astype(int) - 1])
savefig('figures/nnbtest.pdf')
def randomMatrixPermutations(mat):
"""Return a matrix with elements of a triangle permuted without saving the degrees.
"""
s = mat.shape
nruter = array(mat)
indices = find(triu(ones(s), k=1))
aleaInd = permutation(indices)
nruter[unravel_index(indices, s)] = nruter[unravel_index(aleaInd, s)]
for i in range(s[0]):
nruter[i:, i] = nruter[i, i:]
return nruter
def initialRand(self, dens, normX=None):
'''Initialize (x, A) with a certain density 'dens'.
'''
#self.x = rand(self.N)
#self.x[self.x < (1.- dens)] = 0
#self.x[self.x >= (1.- dens)] = 1
#if normX != None:
#self.x *= normX / norm(self.x)
#self.A = self.thresholding(self.x, dens)
self.A = zeros(self.N)
self.A[permutation(self.N)[range(int(dens * self.N))]] = 1.
self.initTheta()
self.x = (arctanh(1.99 *
(self.A - 0.5)) / self.G + self.theta) / self.P
def loadMatlabData(filename):
"""
Load data from csv file and divide it into parts for
training, cross validation and test.
"""
# Load the training data
print "Loading training data..."
data = loadmat(filename)
X = data['X']
y = data['y']
# Randomly select 100 datapoints to display
sel = permutation(X.shape[0])
random_columns = sel[0:100]
displayData(X[random_columns,:])
return X,y
def main(k):
XC = loadtxt('data/iris.data', delimiter=',', dtype=float,
converters={4: cnvt})
ind = arange(150) # indices into the dataset
ind = permutation(ind) # random permutation
L = ind[0:90] # learning set indices
T = ind[90:] # test set indices
# Learning Set
# The k-NNb classifier is called for given k.
X = transpose(XC[L, 0:4])
nnc = NNb(X, XC[L, -1], k)
# Classification of Test Set
c = zeros(len(T))
for i in range(len(T)):
c[i] = nnc.classify(XC[T[i], 0:4])
# Confusion Matrix
CM = zeros((3, 3))
for i in range(3):
for j in range(3):
CM[i, j] = sum(logical_and(XC[T, 4] == (i + 1), c == (j + 1)))
print(CM)
# Plot Test Set
plt.figure(1)
color = array([[1, 0, 0], [0, 1, 0], [0, 0, 1]])
for i in range(4):
for j in range(4):
plt.subplot(4, 4, 4 * i + j + 1)
if i == j:
continue
plt.scatter(XC[T, i], XC[T, j], s=100, marker='s',
edgecolor=color[XC[T, 4].astype(int) - 1],
facecolor=[1, 1, 1] * len(T))
plt.scatter(XC[T, i], XC[T, j], s=30, marker='+',
edgecolor=color[c.astype(int) - 1])
plt.savefig('lab_42.png')
def __init__(self,args):
self.defaults={'fg':3000,
'bg':300,
'pre':500,
'dur':100,
'post':1000,
'reps':20,
'trigger':1,
'res':25,
'box_sz':10,
'sz':250,
'analysis':'spike',
'slope':1,
'respDelay':100,
'box_order':arange(15**2)[permutation(15**2).astype(int)]}
#self.paramsOrder=['fg','bg','sz','res','box_sz','box_order','reps','pre','dur','post','trigger','analysis','slope','respDelay']
self.paramsOrder=['fg','bg','sz','res','box_sz','reps','pre','dur','post','analysis','respDelay']
self.initOther(args)
self.calMod=calibrationModule(self.defaults, self.paramsOrder)
self.calMod.initCalibratedBG()
self.calMod.initCalibratedFG(['sz', 'res'])
self.initParams()
self.protName='squareboardLineCalibrated'
# Get N nearest neighbors
minindex = np.argsort(dsq)[0:self.k]
# Group sum by value
return Counter(self.c[minindex]).most_common()[0][0]
def cnvt(s):
tab = {'Iris-setosa':1.0, 'Iris-versicolor':2.0, 'Iris-virginica':3.0}
if tab.has_key(s):
return tab[s]
else:
return -1.0
XC = loadtxt('data.csv', delimiter=',', dtype=float, converters={4: cnvt})
ind = arange(150) # indices into the dataset
ind = permutation(ind) # random permutation
L = ind[0:90] # learning set indices
T = ind[90:] # test set indices
# Learning Set
X = transpose(XC[L,0:4])
nnc = NNb(X, XC[L,-1])
# Classification of Test Set
c = zeros(len(T))
for i in arange(len(T)):
c[i] = nnc.classify(XC[T[i],0:4])
# Confusion Matrix
CM = zeros((3,3))
for i in range(3):
def getTestParamDict(symbol):
if symbol==key._1:
d={'type':'vsync_test',
'fg':1,
'bg':0.5,
'pre':.5,
'dur':5.,
'post':.5,
'reps':1,
'trigger':0}
elif symbol==key._2:
d={'type':'checkerboard',
'fg':1.0,
'bg':0.5,
'pre':.5,
'dur':5.,
'post':.5,
'sz':400,
'res':array([20,1]),
'mu':0.50,
'std':0.2,
'reps':1,
'trigger':0}
elif symbol==key._3:
d={'type':'squareboard',
'fg':1.0,
'bg':0.2,
'pre':.1,
'dur':.2,
'post':.1,
'radius':50,
'squarenum':permutation(16)+1,
'sz':400,
'res':4,
'reps':1,
'trigger':0}
elif symbol==key._4:
d={'type':'sectors',
'fg':1.0,
'bg':0.5,
'pre':.1,
'dur':.2,
'post':.1,
'radius':100,
'orientation':array((0,90)),
'width':45,
'reps':3,
'trigger':0}
elif symbol==key._5:
d={'type':'fullfieldnoise',
'fg':1.0,
'bg':0.5,
'pre':.2,
'dur':2.,
'post':.2,
'mu':0.50,
'std':0.3,
'reps':2,
'res':3,
'trigger':0}
elif symbol==key._6:
d={'type':'calibration',
'fg':1.0,
'bg':0.5,
'pre':.1,
'dur':.2,
'post':.1,
'intensity':arange(10)/10.,
'reps':1,
'trigger':0}
else:
d=None
return d
