def generate_all_masks(all_It, params_R_M, params_C_M, params_D_M):
all_Mn, all_Mm, all_Ml = [], [], []
for E1,E2,fraction_unknown in params_R_M:
Mn = generate_M(all_It[E1],all_It[E2],fraction_unknown)
all_Mn.append(Mn)
for E1,fraction_unknown in params_C_M:
Mm = generate_M(all_It[E1],all_It[E1],fraction_unknown)
all_Mm.append(Mm)
for E1,J,fraction_unknown in params_D_M:
Ml = generate_M(all_It[E1],J,fraction_unknown)
all_Ml.append(Ml)
return all_Mn, all_Mm, all_Ml
def try_generate_M(I,J,fraction_unknown,attempts):
for attempt in range(1,attempts+1):
try:
M = generate_M(I,J,fraction_unknown)
sums_columns = M.sum(axis=0)
sums_rows = M.sum(axis=1)
for i,c in enumerate(sums_rows):
assert c != 0, "Fully unobserved row in M, row %s. Fraction %s." % (i,fraction_unknown)
for j,c in enumerate(sums_columns):
assert c != 0, "Fully unobserved column in M, column %s. Fraction %s." % (j,fraction_unknown)
print "Took %s attempts to generate M." % attempt
return M
except AssertionError:
pass
raise Exception("Tried to generate M %s times, with I=%s, J=%s, fraction=%s, but failed." % (attempts,I,J,fraction_unknown))
def try_generate_M(I, J, fraction_unknown, attempts):
for attempt in range(1, attempts + 1):
try:
M = generate_M(I, J, fraction_unknown)
sums_columns = M.sum(axis=0)
sums_rows = M.sum(axis=1)
for i, c in enumerate(sums_rows):
assert c != 0, "Fully unobserved row in M, row %s. Fraction %s." % (
i, fraction_unknown)
for j, c in enumerate(sums_columns):
assert c != 0, "Fully unobserved column in M, column %s. Fraction %s." % (
j, fraction_unknown)
print "Took %s attempts to generate M." % attempt
return M
except AssertionError:
pass
raise Exception(
"Tried to generate M %s times, with I=%s, J=%s, fraction=%s, but failed."
% (attempts, I, J, fraction_unknown))
R = numpy.dot(F,numpy.dot(S,G.T))
iterations = 50
init_S = 'random'
init_FG = 'kmeans'
I, J, K, L = 20, 20, 3, 2
fraction_unknown = 0.6
alpha, beta = 1., 1.
lambdaF = numpy.ones((I,K))*1
lambdaS = numpy.ones((K,L))/50
lambdaG = numpy.ones((J,L))*1
priors = { 'alpha':alpha, 'beta':beta, 'lambdaF':lambdaF, 'lambdaS':lambdaS, 'lambdaG':lambdaG }
M = generate_M(I,J,fraction_unknown)
M_test = calc_inverse_M(M)
# Run the Gibbs sampler
BNMTF = bnmtf_vb(R,M,K,L,priors)
BNMTF.initialise(init_S=init_S,init_FG=init_FG)
BNMTF.run(iterations)
# Also measure the performances
performances = BNMTF.predict(M_test)
print performances
# Plot the tau expectation values to check convergence
plt.plot(BNMTF.all_exp_tau)
print "F: %s." % BNMTF.expF
I, J, K, L = 20, 20, 3, 2
fraction_unknown = 0.6
alpha, beta = 1., 1.
lambdaF = numpy.ones((I, K)) * 1
lambdaS = numpy.ones((K, L)) / 50
lambdaG = numpy.ones((J, L)) * 1
priors = {
'alpha': alpha,
'beta': beta,
'lambdaF': lambdaF,
'lambdaS': lambdaS,
'lambdaG': lambdaG
}
M = generate_M(I, J, fraction_unknown)
M_test = calc_inverse_M(M)
# Run the Gibbs sampler
BNMTF = bnmtf_vb(R, M, K, L, priors)
BNMTF.initialise(init_S=init_S, init_FG=init_FG)
BNMTF.run(iterations)
# Also measure the performances
performances = BNMTF.predict(M_test)
print performances
# Plot the tau expectation values to check convergence
plt.plot(BNMTF.all_exp_tau)
print "F: %s." % BNMTF.expF
