osl_dynamics.models.dynemo

Dynamic Network Modes (DyNeMo).

See the documentation for a description of this model.

Classes

Config	Settings for DyNeMo.
Model	DyNeMo model class.

Module Contents

class osl_dynamics.models.dynemo.Config

Bases: osl_dynamics.models.mod_base.BaseModelConfig, osl_dynamics.models.inf_mod_base.VariationalInferenceModelConfig

Settings for DyNeMo.

Parameters:

• model_name (str) – Model name.

• n_modes (int) – Number of modes.

• n_channels (int) – Number of channels.

• sequence_length (int) – Length of sequence passed to the inference network and generative model.

• inference_rnn (str) – RNN to use, either 'gru' or 'lstm'.

• inference_n_layers (int) – Number of layers.

• inference_n_units (int) – Number of units.

• inference_normalization (str) – Type of normalization to use. Either None, 'batch' or 'layer'.

• inference_activation (str) – Type of activation to use after normalization and before dropout. E.g. 'relu', 'elu', etc.

• inference_dropout (float) – Dropout rate.

• inference_regularizer (str) – Regularizer.

• model_rnn (str) – RNN to use, either 'gru' or 'lstm'.

• model_n_layers (int) – Number of layers.

• model_n_units (int) – Number of units.

• model_normalization (str) – Type of normalization to use. Either None, 'batch' or 'layer'.

• model_activation (str) – Type of activation to use after normalization and before dropout. E.g. 'relu', 'elu', etc.

• model_dropout (float) – Dropout rate.

• model_regularizer (str) – Regularizer.

• learn_alpha_temperature (bool) – Should we learn alpha_temperature?

• initial_alpha_temperature (float) – Initial value for alpha_temperature.

• learn_means (bool) – Should we make the mean vectors for each mode trainable?

• learn_covariances (bool) – Should we make the covariance matrix for each mode trainable?

• initial_means (np.ndarray) – Initialisation for mean vectors.

• initial_covariances (np.ndarray) – Initialisation for mode covariances. If diagonal_covariances=True and full matrices are passed, the diagonal is extracted.

• covariances_epsilon (float) – Error added to mode covariances for numerical stability.

• diagonal_covariances (bool) – Should we learn diagonal mode covariances?

• means_regularizer (tf.keras.regularizers.Regularizer) – Regularizer for mean vectors.

• covariances_regularizer (tf.keras.regularizers.Regularizer) – Regularizer for covariance matrices.

• do_kl_annealing (bool) – Should we use KL annealing during training?

• kl_annealing_curve (str) – Type of KL annealing curve. Either 'linear' or 'tanh'.

• kl_annealing_sharpness (float) – Parameter to control the shape of the annealing curve if kl_annealing_curve='tanh'.

• n_kl_annealing_epochs (int) – Number of epochs to perform KL annealing.

• init_method (str) – Initialization method to use. Defaults to ‘random_subset’.

• n_init (int) – Number of initializations. Defaults to 5.

• n_init_epochs (int) – Number of epochs for each initialization. Defaults to 2.

• init_take (float) – Fraction of dataset to use in the initialization. Defaults to 1.0.

• batch_size (int) – Mini-batch size.

• learning_rate (float) – Learning rate.

• lr_decay (float) – Decay for learning rate. Default is 0.1. We use lr = learning_rate * exp(-lr_decay * epoch).

• gradient_clip (float) – Value to clip gradients by. This is the clipnorm argument passed to the Keras optimizer. Cannot be used if multi_gpu=True.

• n_epochs (int) – Number of training epochs.

• optimizer (str or tf.keras.optimizers.Optimizer) – Optimizer to use. 'adam' is recommended.

• loss_calc (str) – How should we collapse the time dimension in the loss? Either 'mean' or 'sum'.

• multi_gpu (bool) – Should be use multiple GPUs for training?

• strategy (str) – Strategy for distributed learning.

• best_of (int) – Number of full training runs to perform. A single run includes its own initialization and fitting from scratch.

model_name: str = 'DyNeMo'

inference_rnn: str = 'lstm'

inference_n_layers: int = 1

inference_n_units: int = None

inference_normalization: str = None

inference_activation: str = None

inference_dropout: float = 0.0

inference_regularizer: str = None

model_rnn: str = 'lstm'

model_n_layers: int = 1

model_n_units: int = None

model_normalization: str = None

model_activation: str = None

model_dropout: float = 0.0

model_regularizer: str = None

learn_means: bool = None

learn_covariances: bool = None

initial_means: numpy.ndarray = None

initial_covariances: numpy.ndarray = None

diagonal_covariances: bool = False

covariances_epsilon: float = None

means_regularizer: tensorflow.keras.regularizers.Regularizer = None

covariances_regularizer: tensorflow.keras.regularizers.Regularizer = None

init_method: str = 'random_subset'

n_init: int = 5

n_init_epochs: int = 2

init_take: float = 1.0

validate_rnn_parameters()

Return type:

None

validate_observation_model_parameters()

Return type:

None

class osl_dynamics.models.dynemo.Model(config)

Bases: osl_dynamics.models.inf_mod_base.VariationalInferenceModelBase

DyNeMo model class.

Parameters:

config (osl_dynamics.models.dynemo.Config)

config_type

build_model()

Builds a keras model.

Return type:

None

get_means()

Get the mode means.

Returns:

means – Mode means.

Return type:

np.ndarary

get_covariances()

Get the mode covariances.

Returns:

covariances – Mode covariances.

Return type:

np.ndarary

get_means_covariances()

Get the mode means and covariances.

This is a wrapper for get_means and get_covariances.

Returns:

• means (np.ndarary) – Mode means.

• covariances (np.ndarray) – Mode covariances.

Return type:

Tuple[numpy.ndarray, numpy.ndarray]

get_observation_model_parameters()

Wrapper for get_means_covariances.

Return type:

Tuple[numpy.ndarray, numpy.ndarray]

set_means(means, update_initializer=True)

Set the mode means.

Parameters:

• means (np.ndarray) – Mode means. Shape is (n_modes, n_channels).

• update_initializer (bool) – Do we want to use the passed means when we re-initialize the model?

Return type:

None

set_covariances(covariances, update_initializer=True)

Set the mode covariances.

Parameters:

• covariances (np.ndarray) – Mode covariances. Shape is (n_modes, n_channels, n_channels).

• update_initializer (bool, optional) – Do we want to use the passed covariances when we re-initialize the model?

Return type:

None

set_means_covariances(means, covariances, update_initializer=True)

This is a wrapper for set_means and set_covariances.

Parameters:

• means (numpy.ndarray)

• covariances (numpy.ndarray)

• update_initializer (bool)

Return type:

None

set_observation_model_parameters(observation_model_parameters, update_initializer=True)

Wrapper for set_means_covariances.

Parameters:

• observation_model_parameters (tuple)

• update_initializer (bool)

Return type:

None

set_regularizers(training_dataset)

Set the means and covariances regularizer based on the training data.

A multivariate normal prior is applied to the mean vectors with mu=0, sigma=diag((range/2)**2). If config.diagonal_covariances=True, a log normal prior is applied to the diagonal of the covariances matrices with mu=0, sigma=sqrt(log(2*range)), otherwise an inverse Wishart prior is applied to the covariances matrices with nu=n_channels-1+0.1 and psi=diag(1/range).

Parameters:

training_dataset (tf.data.Dataset or osl_dynamics.data.Data) – Training dataset.

Return type:

None

sample_alpha(n_samples, theta=None)

Uses the model RNN to sample mode mixing factors, alpha.

Parameters:

• n_samples (int) – Number of samples to take.

• theta (np.ndarray, optional) – Normalized logits to initialise the sampling with. Shape must be (sequence_length, n_modes).

Returns:

alpha – Sampled alpha.

Return type:

np.ndarray

fine_tuning(training_data, n_epochs=None, learning_rate=None, store_dir='tmp')

Fine tuning the model for each session.

Here, we train the inference RNN and observation model with the model RNN fixed held fixed at the group-level.

Parameters:

• training_data (osl_dynamics.data.Data) – Training dataset.

• n_epochs (int, optional) – Number of epochs to train for. Defaults to the value in the config used to create the model.

• learning_rate (float, optional) – Learning rate. Defaults to the value in the config used to create the model.

• store_dir (str, optional) – Directory to temporarily store the model in.

Returns:

• alpha (list of np.ndarray) – Session-specific mixing coefficients. Each element has shape (n_samples, n_modes).

• means (np.ndarray) – Session-specific means. Shape is (n_sessions, n_modes, n_channels).

• covariances (np.ndarray) – Session-specific covariances. Shape is (n_sessions, n_modes, n_channels, n_channels).

Return type:

Tuple[List[numpy.ndarray], numpy.ndarray, numpy.ndarray]

dual_estimation(training_data, n_epochs=None, learning_rate=None, store_dir='tmp')

Dual estimation to get the session-specific observation model parameters.

Here, we train the observation model parameters (mode means and covariances) with the inference RNN and model RNN held fixed at the group-level.

Parameters:

• training_data (osl_dynamics.data.Data or list of tf.data.Dataset) – Training dataset.

• n_epochs (int, optional) – Number of epochs to train for. Defaults to the value in the config used to create the model.

• learning_rate (float, optional) – Learning rate. Defaults to the value in the config used to create the model.

• store_dir (str, optional) – Directory to temporarily store the model in.

Returns:

• means (np.ndarray) – Session-specific means. Shape is (n_sessions, n_modes, n_channels).

• covariances (np.ndarray) – Session-specific covariances. Shape is (n_sessions, n_modes, n_channels, n_channels).

Return type:

Tuple[numpy.ndarray, numpy.ndarray]