def test_lda_recalculate_beta():
K = 2
W = 3
doc0 = [(0,3), (1,1)]
doc1 = [(1,3), (2,2), (0,1)]
text = [doc0,doc1]
text = [doc0,doc1]
beta = np.empty((K,W), dtype=float)
out = beta.copy()
phi0 = np.zeros((sum([d[1] for d in doc0]), K))
# two to topic one (word 0)
# two to topic two (word 1)
phi0[0][0] = 1
phi0[1][0] = 1
phi0[2][1] = 1
phi0[3][1] = 1
phi1 = np.zeros((sum([d[1] for d in doc1]), K))
phi1[0][1] = 1
phi1[1][1] = 1
phi1[2][1] = 1
phi1[3][0] = 1
phi1[4][1] = 1
phi1[5][0] = 1
phi = [phi0, phi1]
answer = np.array([[0.75, 0.0, 0.25],
[1.0/6, 2.0/3, 1.0/6]])
assert out.shape == (2,3)
lm.lda_recalculate_beta(text, out, phi)
assert out.shape == (2,3)
assert not same(beta, out)
assert same(out, answer)
# now test on docarray
out = beta.copy()
assert out.shape == (2,3)
lm.lda_recalculate_beta([lm.doc_to_array(t) for t in text], out, phi)
assert out.shape == (2,3)
assert not same(beta, out)
assert same(out, answer)
# test log space
log_out = np.log(out)
log_phi = [np.log(p) for p in phi]
assert log_out.shape == (2,3)
lm.lda_recalculate_log_beta(text, log_out, log_phi)
assert log_out.shape == (2,3)
assert not same(beta, np.exp(log_out))
assert same(np.exp(log_out), answer)
def partial_slda_m_step(var):
### M-step: ###
print 'updating betas..'
# update betaD for documents first
topiclib.lda_recalculate_beta(var.documents, var.beta, var.phi)
print 'eta sigma...'
# update response variable gaussian global parameters
var.sigma_squared = topiclib.partial_slda_recalculate_eta_sigma(var.eta, var.y, var.phi)
def tlc_m_step(var):
### M-step: ###
print 'updating betas..'
Ku, Ks, Kb = var.Ku, var.Ks, var.Kb
print 'update unlabeled document topics..'
dc = var.documents + var.comments
#phi_dc = var.phiD + [p[:,:Ku] for p in var.phiC]
phi_dc = [d*var.document_multiplier for d in var.phiD] + [var.comment_multiplier*p[:,:Ku] for p in var.phiC]
topiclib.lda_recalculate_beta(dc, var.beta[:Ku], phi_dc)
print 'update sentiment topics...'
cl = var.comments + var.labeled
phi_cl = [var.comment_multiplier*p[:,-Ks:] for p in var.phiC] + [var.labeled_multiplier*p[:,:Ks] for p in var.phiL]
topiclib.lda_recalculate_beta(cl, var.beta[Ku:Ku+Ks], phi_cl)
print 'update background topics...'
lb = var.labeled + var.background
phi_lb = [var.labeled_multiplier*p[:,-Kb:] for p in var.phiL] + [var.background_multiplier*b for b in var.phiB]
topiclib.lda_recalculate_beta(lb, var.beta[-Kb:], phi_lb)
print 'eta sigma...'
# update response variable gaussian global parameters
var.sigma_squared = topiclib.partial_slda_recalculate_eta_sigma(var.eta, var.y, var.phiL)
def lda_m_step(var):
print 'updating betas..'
topiclib.lda_recalculate_beta(var.documents, var.beta, var.phi)
