ed.BiGANInference

Class BiGANInference

Inherits From: GANInference

Aliases:

  • Class ed.BiGANInference
  • Class ed.inferences.BiGANInference

Defined in edward/inferences/bigan_inference.py.

Adversarially Learned Inference (Dumoulin et al., 2017) or Bidirectional Generative Adversarial Networks (Donahue et al., 2017) for joint learning of generator and inference networks.

Works for the class of implicit (and differentiable) probabilistic models. These models do not require a tractable density and assume only a program that generates samples.

Notes

BiGANInference matches a mapping from data to latent variables and a mapping from latent variables to data through a joint discriminator.

In building the computation graph for inference, the discriminator's parameters can be accessed with the variable scope "Disc". In building the computation graph for inference, the encoder and decoder parameters can be accessed with the variable scope "Gen".

Examples

with tf.variable_scope("Gen"):
  xf = gen_data(z_ph)
  zf = gen_latent(x_ph)
inference = ed.BiGANInference({z_ph: zf}, {xf: x_ph}, discriminator)

Methods

init

__init__(
    latent_vars,
    data,
    discriminator
)

build_loss_and_gradients

build_loss_and_gradients(var_list)

finalize

finalize()

Function to call after convergence.

initialize

initialize(
    optimizer=None,
    optimizer_d=None,
    global_step=None,
    global_step_d=None,
    var_list=None,
    *args,
    **kwargs
)

Initialize inference algorithm. It initializes hyperparameters and builds ops for the algorithm's computation graph.

Args:

  • optimizer: str or tf.train.Optimizer, optional. A TensorFlow optimizer, to use for optimizing the generator objective. Alternatively, one can pass in the name of a TensorFlow optimizer, and default parameters for the optimizer will be used.
  • optimizer_d: str or tf.train.Optimizer, optional. A TensorFlow optimizer, to use for optimizing the discriminator objective. Alternatively, one can pass in the name of a TensorFlow optimizer, and default parameters for the optimizer will be used.
  • global_step: tf.Variable, optional. Optional Variable to increment by one after the variables for the generator have been updated. See tf.train.Optimizer.apply_gradients.
  • global_step_d: tf.Variable, optional. Optional Variable to increment by one after the variables for the discriminator have been updated. See tf.train.Optimizer.apply_gradients.
  • var_list: list of tf.Variable, optional. List of TensorFlow variables to optimize over (in the generative model). Default is all trainable variables that latent_vars and data depend on.
print_progress(info_dict)

Print progress to output.

run

run(
    variables=None,
    use_coordinator=True,
    *args,
    **kwargs
)

A simple wrapper to run inference.

  1. Initialize algorithm via initialize.
  2. (Optional) Build a TensorFlow summary writer for TensorBoard.
  3. (Optional) Initialize TensorFlow variables.
  4. (Optional) Start queue runners.
  5. Run update for self.n_iter iterations.
  6. While running, print_progress.
  7. Finalize algorithm via finalize.
  8. (Optional) Stop queue runners.

To customize the way inference is run, run these steps individually.

Args:

  • variables: list, optional. A list of TensorFlow variables to initialize during inference. Default is to initialize all variables (this includes reinitializing variables that were already initialized). To avoid initializing any variables, pass in an empty list.
  • use_coordinator: bool, optional. Whether to start and stop queue runners during inference using a TensorFlow coordinator. For example, queue runners are necessary for batch training with file readers. *args, **kwargs: Passed into initialize.

update

update(
    feed_dict=None,
    variables=None
)

Run one iteration of optimization.

Args:

  • feed_dict: dict, optional. Feed dictionary for a TensorFlow session run. It is used to feed placeholders that are not fed during initialization.
  • variables: str, optional. Which set of variables to update. Either "Disc" or "Gen". Default is both.

Returns:

dict. Dictionary of algorithm-specific information. In this case, the iteration number and generative and discriminative losses.

Notes

The outputted iteration number is the total number of calls to update. Each update may include updating only a subset of parameters.