def testDictionary(NFactors, K, lam, showDict=True):
import spams
from VideoTools import loadImageIOVideo
import time
(I, IDims) = loadImageIOVideo("jumpingjacks2men.ogg")
#Step 1: Build dictionary based on first frame
F1 = np.reshape(I[0, :], IDims)
P = getColorPatchesImageSet([F1], K, getPatches)
ShapeBefore = P.shape[1::]
P = np.reshape(P, [P.shape[0] * P.shape[1] * P.shape[2], P.shape[3]])
print("Doing NNSC for %i factors on %i patches..." %
(NFactors, P.shape[0]))
tic = time.time()
U = spams.nnsc(P.T,
lambda1=lam,
return_lasso=False,
K=NFactors,
numThreads=8)
print("Elapsed Time: %g" % (time.time() - tic))
if showDict:
k = int(np.sqrt(NFactors))
for i in range(k * k):
plt.subplot(k, k, i + 1)
p = unwrapColorPatch(U[:, i], K)
p = p / np.max(p)
plt.imshow(p, interpolation='none')
plt.axis('off')
plt.savefig("Dict_%i_%i_%.3g.png" % (NFactors, K, lam),
bbox_inches='tight')
#Step 2: Represent another frame in the constructed dictionary
F2 = np.reshape(I[10, :], IDims)
P2 = getColorPatchesImageSet([F2], K, getPatches)
ShapeBefore = P2.shape[1::]
P2 = np.reshape(P2, [P2.shape[0] * P2.shape[1] * P2.shape[2], P2.shape[3]])
print("Solving for factors")
tic = time.time()
Alpha = spams.lasso(P2.T, U, lambda1=lam, pos=True, numThreads=8)
print("Elapsed Time: %g" % (time.time() - tic))
P2New = np.reshape((U * Alpha).T, ShapeBefore)
F2Reconstructed = recombineColorPatches(P2New, K, F2.shape[0], F2.shape[1])
writeImage(F2, "Orig.png")
writeImage(F2Reconstructed,
"Reconstructed_%i_%i_%.3g.png" % (NFactors, K, lam))
print("%i Factors, Avg nonzero elems: %.3g" %
(NFactors,
NFactors * Alpha.size / float(Alpha.shape[0] * Alpha.shape[1])))
def online_NMF(self,n_components=30,method='nnsc',lambda1=100,iterations=-5,batchsize=512,model=None,**kwargs):
""" Method performing online matrix factorization and using the spams (http://spams-devel.gforge.inria.fr/doc-python/html/index.html) package from Inria.
Implements bith the nmf and nnsc methods
Parameters
----------
n_components: int
method: 'nnsc' or 'nmf' (see http://spams-devel.gforge.inria.fr/doc-python/html/index.html)
lambda1: see http://spams-devel.gforge.inria.fr/doc-python/html/index.html
iterations: see http://spams-devel.gforge.inria.fr/doc-python/html/index.html
batchsize: see http://spams-devel.gforge.inria.fr/doc-python/html/index.html
model: see http://spams-devel.gforge.inria.fr/doc-python/html/index.html
**kwargs: more arguments to be passed to nmf or nnsc
Return:
-------
time_comps
space_comps
"""
try:
import spams
except:
print("You need to install the SPAMS package")
raise
T,d1,d2=np.shape(self)
d=d1*d2
X=np.asfortranarray(np.reshape(self,[T,d],order='F'))
if method == 'nmf':
(time_comps,V) = spams.nmf(X,return_lasso= True ,K = n_components,numThreads=4,iter = iterations,**kwargs)
elif method == 'nnsc':
(time_comps,V) = spams.nnsc(X,return_lasso=True,K=n_components, lambda1 = lambda1,iter = iterations, model = model, **kwargs)
else:
raise Exception('Method unknown')
space_comps=[]
for idx,mm in enumerate(V):
space_comps.append(np.reshape(mm.todense(),(d1,d2),order='F'))
return time_comps,np.array(space_comps)
else:
assign = [
np.loadtxt(x, ndmin=2) for x in files.glob('*mutation_assignment*')
]
ccms = [np.dot(x, x.T) for x in assign]
CCM = sum(ccms)
counts = [sum(x.T)[:, None] for x in assign]
counting = sum([np.dot(x, x.T) for x in counts])
X = np.round(CCM / (1e-12 + counting), 3)
X = np.asfortranarray(X / np.tile(np.sqrt((X * X).sum(axis=0)),
(X.shape[0], 1)),
dtype=float)
# SPAMS
(U, V) = spams.nnsc(X,
return_lasso=True,
K=int(NCL),
numThreads=-1,
iter=iters,
lambda1=0)
lab = list(map(np.argmax, U))
label = np.array(lab)
labels = np.unique(label)
NCL = len(labels)
ssm = [np.sum(label == x) for x in labels]
tmp = [np.mean(phi[label == x]) for x in labels]
suma = np.zeros((NCL, 3))
suma[:, 0] = labels
suma[:, 1] = ssm
suma[:, 2] = tmp
suma = suma[np.argsort(suma[:, 2]), ]
refine = False
if NCL > 1 and any((suma[1:, 2] - suma[:-1, 2]) < 0.05):
#%%
mcr = cm.load(fname)
ycr = mcr.to_2D()
ycr = ycr - ycr.min()
mcr_lc = local_correlations_movie_offline(fname)
ycr_lc = mcr_lc.to_2D()
#%%
D_lc,tr_lc = spams.nmf(np.asfortranarray(ycr_lc.T), K=2, return_lasso=True)
plt.figure();plt.plot(tr_lc.T.toarray())
plt.figure();plt.imshow(D_lc[:,0].reshape(mcr.shape[1:], order='F'), cmap='gray')
#%%
D,tr = spams.trainDL(np.asfortranarray(ycr.T), K=2, D=D_lc, lambda1=0)
#%%
D,tr = spams.nnsc(np.asfortranarray(ycr.T), K=2, return_lasso=True, lambda1=0)
#%%
D,tr = spams.nmf(np.asfortranarray(np.abs(ycr.T)), K=2, return_lasso=True)
#%%
D,tr = spams.nnsc(np.asfortranarray(ycr.T), K=2, return_lasso=True, lambda1=1)
#%%
plt.figure();plt.plot(tr.T.toarray())
plt.figure();plt.imshow(D[:,1].reshape(mcr.shape[1:], order='F'), cmap='gray');plt.title('comp1')
plt.figure();plt.imshow(D[:,0].reshape(mcr.shape[1:], order='F'), cmap='gray');plt.title('comp0')
#%%
plt.figure();plt.plot(tr.T.toarray())
plt.figure();plt.imshow(D[:,1].reshape(mcr.shape[1:], order='F'), cmap='gray')
yr = np.reshape(m1, [T, d], order='F').T
X = np.asfortranarray(yr, dtype=myfloat)
########## FIRST EXPERIMENT ###########
tic = time.time()
# if you want to use NMF uncomment the following line
#(U,V) = spams.nmf(X,return_lasso= True,K = n_components,iter = -5)
# regularizer on space components
#spams.trainDL(X,return_model = True,D=np.asfortranarray(V.todense().T),posAlpha=True,posD=True,modeD=3,gamma1=.001,lambda1=3000,lambda2=0,mode=spams.spams_wrap.PENALTY,iter=-5)
if not use_pixels_as_basis:
# (D,model)=spams.trainDL(X,return_model = True,K=n_components,posAlpha=True,posD=True,modeD=3,gamma1=.001,lambda1=2000,lambda2=0,mode=spams.spams_wrap.PENALTY,iter=-5)
# U=model['B']
# V=model['A']
(U, V) = spams.nnsc(X, return_lasso=True,
K=n_components, lambda1=1000, iter=-5)
else:
(U, V) = spams.nnsc(X, return_lasso=True,
K=n_components, lambda1=1000, iter=-5)
#(U,V) = spams.nnsc(np.asfortranarray(yr.T,dtype = myfloat),return_lasso=True,K=15,lambda1=100,iter=-5)
# model=dict()
# model['A']=np.asfortranarray(V.todense().T)
# model['B']=np.asfortranarray(U)
# model['iter']=100
#(D,model) = spams.trainDL(X,return_model = True,D=np.asfortranarray(V.todense().T),posAlpha=True,posD=True,modeD=3,gamma1=.001,lambda1=3000,lambda2=0,mode=spams.spams_wrap.PENALTY,iter=-5)
if use_pixels_as_basis:
comp = V.todense()
def online_NMF(self,
n_components=30,
method='nnsc',
lambda1=100,
iterations=-5,
batchsize=512,
model=None,
**kwargs):
""" Method performing online matrix factorization and using the spams (http://spams-devel.gforge.inria.fr/doc-python/html/index.html) package from Inria.
Implements bith the nmf and nnsc methods
Parameters
----------
n_components: int
method: 'nnsc' or 'nmf' (see http://spams-devel.gforge.inria.fr/doc-python/html/index.html)
lambda1: see http://spams-devel.gforge.inria.fr/doc-python/html/index.html
iterations: see http://spams-devel.gforge.inria.fr/doc-python/html/index.html
batchsize: see http://spams-devel.gforge.inria.fr/doc-python/html/index.html
model: see http://spams-devel.gforge.inria.fr/doc-python/html/index.html
**kwargs: more arguments to be passed to nmf or nnsc
Return:
-------
time_comps
space_comps
"""
try:
import spams
except:
print "You need to install the SPAMS package"
raise
T, d1, d2 = np.shape(self)
d = d1 * d2
X = np.asfortranarray(np.reshape(self, [T, d], order='F'))
if method == 'nmf':
(time_comps, V) = spams.nmf(X,
return_lasso=True,
K=n_components,
numThreads=4,
iter=iterations,
**kwargs)
elif method == 'nnsc':
(time_comps, V) = spams.nnsc(X,
return_lasso=True,
K=n_components,
lambda1=lambda1,
iter=iterations,
model=model,
**kwargs)
else:
raise Exception('Method unknown')
space_comps = []
for idx, mm in enumerate(V):
space_comps.append(np.reshape(mm.todense(), (d1, d2), order='F'))
return time_comps, np.array(space_comps)
fout.write(line)
foutD.write(line)
line = fin.readline()
X = [float(a) for a in line.split(",")[0:-1]]
X = np.reshape(X, [len(X)/12, 12])
ND = X.shape[0] - BeatsPerBlock + 1
if ND <= 0:
fout.write("\n")
foutD.write("\n")
print "ND = 0"
continue
print "(NPixels, ND) = (%i, %i)"%(NPixels, ND)
OutDs = np.zeros((NPixels, ND))
for i in range(ND):
x = X[i:i+BeatsPerBlock, :]
D = getSSM(x, DPixels)
OutDs[:, i] = D[I < J]
Dict = spams.nnsc(np.asfortranarray(OutDs), lambda1 = 1, K = NDictElems)
#Output colomn-major
Dict = (Dict.T).flatten()
OutDs = (OutDs.T).flatten()
np.savetxt(fout, OutDs, fmt="%g", newline=",")
fout.write("\n")
np.savetxt(foutD, Dict, fmt="%g", newline=",")
foutD.write("\n")
line = fin.readline()
idx += 1
fin.close()
fout.close()
foutD.close()