nvis = dim
)
#generate the noise samples
noise_func = function([], noise_distribution.random_design_matrix(X))
Y = []
for i in xrange(noise_per_clean):
Y.append(sharedX(noise_func()))
#Get the objective function
nce = DNCE(noise_distribution)
J = nce(model,X,Y)
accs = []
for Y_i in Y:
pos_prob = 1./(1.+T.exp(model.free_energy(X)-model.free_energy(Y_i)))
acc = (pos_prob > .5).mean()
accs.append(acc)
acc = sum(accs) / float(len(accs))
print '\tinit accuracy ',function([],acc)()
#Minimize the objective function with batch gradient descent
minimizer = BatchGradientDescent( objective = J,
params = model.get_params(),
param_constrainers = [ model.censor_updates ])
print '\tinit obj:',minimizer.obj()
#minimizer.verbose = True
minimizer.minimize()
print '\tfinal obj:',minimizer.obj()
for i in xrange(noise_per_clean):
Y.append(sharedX(noise_func()))
#Get the objective function
nce = NCE(
DiagonalMND(nvis=dim,
init_beta=beta,
init_mu=0.,
min_beta=beta,
max_beta=beta), -1)
J = nce(model, X, T.concatenate(Y, axis=0))
accs = []
for Y_i in Y:
pos_prob = 1. / (
1. + T.exp(model.free_energy(X) - model.free_energy(Y_i)))
acc = (pos_prob > .5).mean()
accs.append(acc)
acc = sum(accs) / float(len(accs))
print '\tinit accuracy ', function([], acc)()
#Minimize the objective function with batch gradient descent
minimizer = BatchGradientDescent(
objective=J,
params=model.get_params(),
param_constrainers=[model.censor_updates])
print '\tinit obj:', minimizer.obj()
#minimizer.verbose = True
minimizer.minimize()
nvis = dim
)
#generate the noise samples
noise_func = function([], noise_distribution.random_design_matrix(X))
Y = []
for i in xrange(noise_per_clean):
Y.append(sharedX(noise_func()))
#Get the objective function
nce = DNCE(noise_distribution)
J = nce(model,X,Y)
accs = []
for Y_i in Y:
pos_prob = 1./(1.+T.exp(model.free_energy(X)-model.free_energy(Y_i)))
acc = (pos_prob > .5).mean()
accs.append(acc)
acc = sum(accs) / float(len(accs))
print '\tinit accuracy ',function([],acc)()
#Minimize the objective function with batch gradient descent
minimizer = BatchGradientDescent( objective = J,
params = model.get_params(),
param_constrainers = [ model.censor_updates ])
print '\tinit obj:',minimizer.obj()
#minimizer.verbose = True
minimizer.minimize()
print '\tfinal obj:',minimizer.obj()