nrows=10,
ncols=10,
figtitle=""):
"""Will open a .pmat .dmat .amat or .vmat file and consider the beginning of each row a imgheight x imgwidth imagette.
These images will be interactively displayed in a nrows x ncols grid of imagettes."""
data = None
if figtitle == "":
figtitle = matfile
if matfile.endswith(".pmat"):
# Use pure python implementation of pmat (faster loading)
from plearn.vmat.PMat import PMat
data = PMat(matfile)
else:
# Use of VMat through the Python-bridge
from plearn.pyext import AutoVMatrix
data = AutoVMatrix(filename=matfile)
showRowsAsImages(data,
img_height=imgheight,
img_width=imgwidth,
nrows=nrows,
ncols=ncols,
figtitle=figtitle)
####################
### main program ###
if __name__ == '__main__':
from plearn.utilities.autoscript import autoscript
autoscript(show_rows_as_images, True)
# library, go to the PLearn Web site at www.plearn.org
# Author: Pascal Vincent
from plearn.plotting.netplot import showRowsAsImages
def show_rows_as_images(matfile, imgheight, imgwidth, nrows=10, ncols=10, figtitle=""):
"""Will open a .pmat .dmat .amat or .vmat file and consider the beginning of each row a imgheight x imgwidth imagette.
These images will be interactively displayed in a nrows x ncols grid of imagettes."""
data = None
if figtitle=="":
figtitle = matfile
if matfile.endswith(".pmat"):
# Use pure python implementation of pmat (faster loading)
from plearn.vmat.PMat import PMat
data = PMat(matfile)
else:
# Use of VMat through the Python-bridge
from plearn.pyext import AutoVMatrix
data = AutoVMatrix(filename=matfile)
showRowsAsImages(data, img_height=imgheight, img_width=imgwidth, nrows=nrows, ncols=ncols, figtitle=figtitle)
####################
### main program ###
if __name__ == '__main__':
from plearn.utilities.autoscript import autoscript
autoscript(show_rows_as_images, True)
for i in range(len(Cd)):
C2[i, i] = Cd[i]
eigvals, eigvecs = eig(C2)
return real(eigvecs.T)
def computeDenoisingFiltersFromCovariance(C, lambd=1e-6, nu=0.10):
C = C + diag(len(C) * [lambd])
Cd = C.diagonal()
C2 = C * (1.0 - nu)
# copy back intial diagonal
for i in range(len(Cd)):
C2[i, i] = Cd[i]
WW = dot(inv(C2), C)
return WW.T
def computeDenoisingEigenFiltersFromCovariance(C, lambd=1e-6, nu=0.10):
WW = computeDenoisingFiltersFromCovariance(C, lambd, nu).T
eigvals, eigvecs = eig(WW)
return real(eigvecs.T)
# return real(inv(eigvecs).T)
####################
### main program ###
if __name__ == "__main__":
from plearn.utilities.autoscript import autoscript
autoscript(computeAndShowFilters, True)
C2 = C*(1.0-nu)
# copy back intial diagonal
for i in range(len(Cd)):
C2[i,i] = Cd[i]
eigvals, eigvecs = eig(C2)
return real(eigvecs.T)
def computeDenoisingFiltersFromCovariance(C, lambd=1e-6, nu=0.10):
C = C+diag(len(C)*[lambd])
Cd = C.diagonal()
C2 = C*(1.0-nu)
# copy back intial diagonal
for i in range(len(Cd)):
C2[i,i] = Cd[i]
WW = dot(inv(C2),C)
return WW.T
def computeDenoisingEigenFiltersFromCovariance(C, lambd=1e-6, nu=0.10):
WW = computeDenoisingFiltersFromCovariance(C, lambd, nu).T
eigvals, eigvecs = eig(WW)
return real(eigvecs.T)
# return real(inv(eigvecs).T)
####################
### main program ###
if __name__ == "__main__":
from plearn.utilities.autoscript import autoscript
autoscript(computeAndShowFilters, True)
# ncomponents=ncomponents,
# constrain_norm_type=constrain_norm_type,
# cov_transformation_type=cov_transformation_type,
# diag_weight = diag_weight,
# diag_nonlinearity = diag_nonlinearity,
# diag_premul = diag_premul,
# offdiag_weight = offdiag_weight,
# offdiag_nonlinearity = offdiag_nonlinearity,
# offdiag_premul = offdiag_premul,
# force_zero_mean = force_zero_mean,
# lr=0.01, nsteps=1, optimizer_nsteps=10)
if __name__ == "__main__":
from plearn.utilities.autoscript import autoscript
helptext = """
OLDEXAMPLE: dcaexperiment.py 123:1 123 2 -2 cov -1 square 1 1 square 1 0"
OLDEXAMPLE: dcaexperiment.py 121:-2 123 4 -2 squaredist 0 exp 1 1 exp -1.6 0"
OLDEXAMPLE: dcaexperiment.py /data/icml07data/mnist_basic/plearn/mnist_basic2_train.pmat 125 400 -2 squaredist 0 exp -1 1 exp -1 0
Ex:
data_set=123:1 seed=1827 ncomponents=2 nonlinearity=none constrain_norm_type=-2 cov_transformation_type=cov diag_add=0. diag_weight=0. diag_nonlinearity=square diag_premul=1.0 offdiag_weight=1.0 offdiag_nonlinearity=exp offdiag_premul=1.0 lr=0.01 nsteps=1 optimizer_nsteps=1 force_zero_mean=False
"""
autoscript(DCAExperiment,True,helptext=helptext)
