Gan

Ok, so I think that the modifications I made here can be merged ...

Two main changes as compared to the master:

1. Added some layers
2. Modified the way variable name and scope is handled in layers (and in SequenceNetwork as well)

Change 1 should by no means affect what has been done until now (on the master). However, change 2 may require a bit of adaptation, especially when defining architectures and layers ...

@tiago.pereira I'll let you check and then we can discuss

Also, I think that this will be irrelevant when we'll go to tf-slim anyway !

bob/learn/tensorflow/layers/ConditionConcat.py

+#!/usr/bin/env python
+# vim: set fileencoding=utf-8 :
+
+import tensorflow as tf
+from .Layer import Layer
+import numpy
+
+import bob.core.log
+import logging
+logger = logging.getLogger("bob.learn")
+
+class ConditionConcat(Layer):
+  """
+  Layer taking a 1d tensor as input and concatenate it with a conditional vector 
+
+  **Parameters**
+
+  name: str
+   The name of the layer
+
+  conditional_dim: int
+   The dimension of the conditional one-hot vector
+
+  use_gpu: bool
+   Store data in the GPU
+
+  """
+
+  def __init__(self, 
+               conditional_dim,
+               name,
+               ):
+
+    super(ConditionConcat, self).__init__(name=name)
+
+    self.conditional_dim = conditional_dim
+    
+    logger.info("+ adding a Condition Concatenation layer ({0}) +".format(name))
+    logger.info("\t conditional dim = {0}".format(conditional_dim))
+
+  def create_variables(self, input_layer, scope=None):
+ 
+    self.input_layer = input_layer
+    logger.info("== registering input in layer {0}".format(self.name))
+
+
+  def get_graph(self, y):
+    
+    batch_size = input_layer.get_shape()[0]
+    yb = tf.reshape(y, [batch_size, 1, 1, self.conditional_dim])
+    return tf.concat([input_layer, y], 1)
+