Module flowcon.flows.autoregressive

Implementations of autoregressive flows.

Classes

class MaskedAutoregressiveFlow (features, hidden_features, num_layers, num_blocks_per_layer, use_residual_blocks=True, use_random_masks=False, use_random_permutations=False, activation=<function relu>, dropout_probability=0.0, batch_norm_within_layers=False, batch_norm_between_layers=False)

An autoregressive flow that uses affine transforms with masking.

Reference:

G. Papamakarios et al., Masked Autoregressive Flow for Density Estimation, Advances in Neural Information Processing Systems, 2017.

Constructor.

Args

transform
A Transform object, it transforms data into noise.
distribution
A AutoregressiveTransform object, the base distribution of the flow that generates the noise.
embedding_net
A nn.Module which has trainable parameters to encode the context (condition). It is trained jointly with the flow.
Expand source code 
class MaskedAutoregressiveFlow(Flow):
    """An autoregressive flow that uses affine transforms with masking.

    Reference:
    > G. Papamakarios et al., Masked Autoregressive Flow for Density Estimation,
    > Advances in Neural Information Processing Systems, 2017.
    """

    def __init__(
        self,
        features,
        hidden_features,
        num_layers,
        num_blocks_per_layer,
        use_residual_blocks=True,
        use_random_masks=False,
        use_random_permutations=False,
        activation=F.relu,
        dropout_probability=0.0,
        batch_norm_within_layers=False,
        batch_norm_between_layers=False,
    ):

        if use_random_permutations:
            permutation_constructor = RandomPermutation
        else:
            permutation_constructor = ReversePermutation

        layers = []
        for _ in range(num_layers):
            layers.append(permutation_constructor(features))
            layers.append(
                MaskedAffineAutoregressiveTransform(
                    features=features,
                    hidden_features=hidden_features,
                    num_blocks=num_blocks_per_layer,
                    use_residual_blocks=use_residual_blocks,
                    random_mask=use_random_masks,
                    activation=activation,
                    dropout_probability=dropout_probability,
                    use_batch_norm=batch_norm_within_layers,
                )
            )
            if batch_norm_between_layers:
                layers.append(BatchNorm(features))

        super().__init__(
            transform=CompositeTransform(layers),
            distribution=StandardNormal([features]),
        )

Ancestors

Class variables

var call_super_init : bool
var dump_patches : bool
var training : bool

Inherited members