def compute_logpx(self, x, y, z):
px_in = tf.reshape(tf.concat([y,z], axis=-1), [-1, self.n_y+self.n_z])
if self.x_dist == 'Gaussian':
mean, log_var = dgm.forwardPassGauss(px_in, self.px_yz, self.n_hid, self.nonlinearity, self.bn, scope='px_yz')
mean, log_var = tf.reshape(mean, [self.mc_samples, -1, self.n_x]), tf.reshape(log_var, [self.mc_samples, -1, self.n_x])
return dgm.gaussianLogDensity(x, mean, log_var)
elif self.x_dist == 'Bernoulli':
logits = dgm.forwardPassCatLogits(px_in, self.px_yz, self.n_hid, self.nonlinearity, self.bn, scope='px_yz')
logits = tf.reshape(logits, [self.mc_samples, -1, self.n_x])
return dgm.bernoulliLogDensity(x, logits)
def compute_logpx(self, x, z):
""" compute the log density of x under p(x|z) """
px_in = tf.reshape(z, [-1, self.n_z])
if self.x_dist == 'Gaussian':
mean, log_var = dgm.forwardPassGauss(px_in, self.px_z, self.n_hid, self.nonlinearity, self.bn, scope='px_z')
mean, log_var = tf.reshape(mean, [self.mc_samples, -1, self.n_x]), tf.reshape(log_var, [self.mc_samples, -1, self.n_x])
return dgm.gaussianLogDensity(x, mean, log_var)
elif self.x_dist == 'Bernoulli':
logits = dgm.forwardPassCatLogits(px_in, self.px_z, self.n_hid, self.nonlinearity, self.bn, scope='px_z')
logits = tf.reshape(logits, [self.mc_samples, -1, self.n_x])
return dgm.bernoulliLogDensity(x, logits)
def build_model(self): self.n_train, self.n = 1,1 self.create_placeholders() if self.y_dist == 'gaussian': self.q = dgm.initGaussBNN(self.n_x, self.n_hid, self.n_y, 'network', initVar=self.initVar, bn=self.bn) self.wTilde = dgm.sampleGaussBNN(self.q, self.n_hid) self.y_m, self.y_lv = dgm.forwardPassGauss(self.x, self.wTilde, self.q, self.n_hid, self.nonlinearity, self.bn, training=True, scope='q', reuse=False) elif self.y_dist == 'categorical': self.q = dgm.initCatBNN(self.n_x, self.n_hid, self.n_y, 'network', initVar=self.initVar, bn=self.bn) self.wTilde = dgm.sampleCatBNN(self.q, self.n_hid) self.y_logits = dgm.forwardPassCatLogits(self.x, self.wTilde, self.q, self.n_hid, self.nonlinearity, self.bn, training=True, scope='q', reuse=False) self.predictions = tf.reduce_mean(self.predict(self.x, 10, training=True),0)
def lowerBound(self, x, y, z, z_m, z_lv, a, qa_m, qa_lv):
""" Helper function for loss computations. Assumes each input is a rank(3) tensor """
pa_in = tf.reshape(tf.concat([y, z], axis=-1),
[-1, self.n_y + self.n_z])
pa_m, pa_lv = dgm.forwardPassGauss(pa_in,
self.pa_yz,
self.n_hid,
self.nonlinearity,
self.bn,
scope='pa_yz')
pa_m, pa_lv = tf.reshape(pa_m,
[self.mc_samples, -1, self.n_a]), tf.reshape(
pa_lv, [self.mc_samples, -1, self.n_a])
l_px = self.compute_logpx(x, y, z, a)
l_py = dgm.multinoulliUniformLogDensity(y)
l_pz = dgm.standardNormalLogDensity(z)
l_pa = dgm.gaussianLogDensity(a, pa_m, pa_lv)
l_qz = dgm.gaussianLogDensity(z, z_m, z_lv)
l_qa = dgm.gaussianLogDensity(a, qa_m, qa_lv)
return tf.reduce_mean(l_px + l_py + l_pz + l_pa - l_qz - l_qa, axis=0)
def compute_logpx(self, x, y, z, a):
""" compute the log density of x under p(x|y,z,a) """
px_in = tf.reshape(tf.concat([y, z, a], axis=-1),
[-1, self.n_y + self.n_z + self.n_a])
if self.x_dist == 'Gaussian':
mean, logVar = dgm.forwardPassGauss(px_in,
self.px_yza,
self.n_hid,
self.nonlinearity,
self.bn,
scope='px_yza')
mean, logVar = tf.reshape(
mean, [self.mc_samples, -1, self.n_x]), tf.reshape(
logVar, [self.mc_samples, -1, self.n_x])
return dgm.gaussianLogDensity(x, mean, logVar)
elif self.x_dist == 'Bernoulli':
logits = dgm.forwardPassCatLogits(px_in,
self.px_yza,
self.n_hid,
self.nonlinearity,
self.bn,
scope='px_yza')
logits = tf.reshape(logits, [self.mc_samples, -1, self.n_x])
return dgm.bernoulliLogDensity(x, logits)
def predictConditionW(self, x, training=True): """ return E[p(y|x, wTilde)] (assumes wTilde~q(W) has been sampled) """ if self.y_dist == 'gaussian': return dgm.forwardPassGauss(x, self.wTilde, self.q, self.n_hid, self.nonlinearity, self.bn, training, scope='q') elif self.y_dist == 'categorical': return dgm.forwardPassCatLogits(x, self.wTilde, self.q, self.n_hid, self.nonlinearity, self.bn, training, scope='q')