def _init_exprs(self):
self.exprs = {
'inpt_mean': T.matrix('inpt_mean'),
'target': T.matrix('target')
}
P = self.parameters
hidden_to_hiddens = [
getattr(P, 'hidden_to_hidden_%i' % i)
for i in range(len(self.n_hiddens) - 1)
]
hidden_biases = [
getattr(P, 'hidden_bias_%i' % i)
for i in range(len(self.n_hiddens))
]
inpt_var = T.zeros_like(self.exprs['inpt_mean']) + self.inpt_var
self.exprs.update(
varprop_mlp.exprs(self.exprs['inpt_mean'], inpt_var,
P.in_to_hidden, hidden_to_hiddens,
P.hidden_to_out, hidden_biases,
[1 for _ in hidden_biases], P.out_bias, 1,
self.hidden_transfers, self.out_transfer,
self.p_dropout_inpt, self.p_dropout_hiddens))
self.exprs['inpt'] = self.exprs['inpt_mean']
self.exprs.update(
varprop_supervised_loss(self.exprs['target'], self.exprs['output'],
self.loss))
def _init_exprs(self):
self.exprs = {
'inpt_mean': T.matrix('inpt_mean'),
'target': T.matrix('target')}
P = self.parameters
hidden_to_hiddens = [getattr(P, 'hidden_to_hidden_%i' % i)
for i in range(len(self.n_hiddens) - 1)]
hidden_biases = [getattr(P, 'hidden_bias_%i' % i)
for i in range(len(self.n_hiddens))]
inpt_var = T.zeros_like(self.exprs['inpt_mean']) + self.inpt_var
self.exprs.update(varprop_mlp.exprs(
self.exprs['inpt_mean'], inpt_var, self.exprs['target'],
P.in_to_hidden,
hidden_to_hiddens,
P.hidden_to_out,
hidden_biases,
[1 for _ in hidden_biases],
P.out_bias,
1,
self.hidden_transfers, self.out_transfer,
self.p_dropout_inpt, self.p_dropout_hiddens))
self.exprs['inpt'] = self.exprs['inpt_mean']
self.exprs.update(varprop_supervised_loss(
self.exprs['target'], self.exprs['output'], self.loss))
def _gen_exprs(self, inpt):
"""Return the expression of the generating model."""
P = self.parameters.gen
n_layers = len(self.n_hiddens_gen)
hidden_to_hiddens = [
getattr(P, 'hidden_to_hidden_%i' % i) for i in range(n_layers - 1)
]
hidden_biases = [
getattr(P, 'hidden_bias_%i' % i) for i in range(n_layers)
]
exprs = vpmlp.exprs(
inpt,
T.zeros_like(inpt),
P.in_to_hidden,
hidden_to_hiddens,
P.hidden_to_out,
hidden_biases,
[1 for _ in hidden_biases],
P.out_bias,
1,
self.gen_transfers,
self.assumptions.statify_visible,
self.p_dropout_inpt,
self.p_dropout_hiddens,
)
return exprs
def _recog_exprs(self, inpt):
"""Return the exprssions of the recognition model."""
P = self.parameters.recog
n_layers = len(self.n_hiddens_recog)
hidden_to_hiddens = [
getattr(P, 'hidden_to_hidden_%i' % i) for i in range(n_layers - 1)
]
hidden_biases = [
getattr(P, 'hidden_bias_%i' % i) for i in range(n_layers)
]
scale_to_interval = lambda x: T.nnet.sigmoid(x) * 0.49 + 0.01
exprs = vpmlp.exprs(
inpt,
T.zeros_like(inpt),
P.in_to_hidden,
hidden_to_hiddens,
P.hidden_to_out,
hidden_biases,
[1 for _ in hidden_biases],
P.out_bias,
1,
self.recog_transfers,
self.assumptions.statify_latent,
scale_to_interval(P.p_dropout.inpt),
[scale_to_interval(i) for i in P.p_dropout.hiddens],
)
return exprs
def _gen_exprs(self, inpt):
"""Return the expression of the generating model."""
P = self.parameters.gen
n_layers = len(self.n_hiddens_gen)
hidden_to_hiddens = [getattr(P, 'hidden_to_hidden_%i' % i)
for i in range(n_layers - 1)]
hidden_biases = [getattr(P, 'hidden_bias_%i' % i)
for i in range(n_layers)]
exprs = vpmlp.exprs(
inpt, T.zeros_like(inpt), P.in_to_hidden, hidden_to_hiddens, P.hidden_to_out,
hidden_biases, [1 for _ in hidden_biases], P.out_bias, 1,
self.gen_transfers, self.assumptions.statify_visible,
self.p_dropout_inpt, self.p_dropout_hiddens,
)
return exprs
def _recog_exprs(self, inpt):
"""Return the exprssions of the recognition model."""
P = self.parameters.recog
n_layers = len(self.n_hiddens_recog)
hidden_to_hiddens = [getattr(P, 'hidden_to_hidden_%i' % i)
for i in range(n_layers - 1)]
hidden_biases = [getattr(P, 'hidden_bias_%i' % i)
for i in range(n_layers)]
scale_to_interval = lambda x: T.nnet.sigmoid(x) * 0.49 + 0.01
exprs = vpmlp.exprs(
inpt, T.zeros_like(inpt), P.in_to_hidden, hidden_to_hiddens, P.hidden_to_out,
hidden_biases, [1 for _ in hidden_biases], P.out_bias, 1,
self.recog_transfers, self.assumptions.statify_latent,
scale_to_interval(P.p_dropout.inpt),
[scale_to_interval(i) for i in P.p_dropout.hiddens],
)
return exprs