torchfilter.train

Reference implementations for training state estimators with learnable parameters.

These are written with a custom model manager for brevity, but can be easily translated to raw PyTorch.

Package Contents

Functions

train_dynamics_recurrent(buddy: fannypack.utils.Buddy, dynamics_model: torchfilter.base.DynamicsModel, dataloader: DataLoader, *, loss_function: str = 'nll', log_interval: int = 10) → None	Trains a dynamics model via backpropagation through time.
train_dynamics_single_step(buddy: fannypack.utils.Buddy, dynamics_model: torchfilter.base.DynamicsModel, dataloader: DataLoader, *, loss_function: str = 'nll', log_interval: int = 10) → None	Optimizes a dynamics model's single-step prediction accuracy. This is roughly
train_filter(buddy: fannypack.utils.Buddy, filter_model: torchfilter.base.Filter, dataloader: DataLoader, initial_covariance: torch.Tensor, *, loss_function: Callable[[torch.Tensor, torch.Tensor], torch.Tensor] = F.mse_loss, log_interval: int = 10, measurement_initialize=False, optimizer_name='train_filter_recurrent') → None	Trains a filter end-to-end via backpropagation through time for 1 epoch over a
train_kalman_filter_measurement(buddy: fannypack.utils.Buddy, measurement_model: torchfilter.base.KalmanFilterMeasurementModel, dataloader: DataLoader, *, loss_function: Callable[[torch.Tensor, torch.Tensor], torch.Tensor] = F.mse_loss, log_interval: int = 10, optimizer_name='train_kalman_filter_measurement') → None	Optimizes a Kalman filter measurement model's prediction accuracy. Minimizes
train_particle_filter_measurement(buddy: fannypack.utils.Buddy, measurement_model: torchfilter.base.ParticleFilterMeasurementModel, dataloader: DataLoader, *, loss_function: Callable[[torch.Tensor, torch.Tensor], torch.Tensor] = F.mse_loss, log_interval: int = 10) → None	Reference implementation for pre-training a particle filter measurement model.
train_virtual_sensor(buddy: fannypack.utils.Buddy, virtual_sensor_model: torchfilter.base.VirtualSensorModel, dataloader: DataLoader, *, loss_function: Callable[[torch.Tensor, torch.Tensor], torch.Tensor] = F.mse_loss, log_interval: int = 10, optimizer_name='train_measurement') → None	Optimizes a virtual sensor model's prediction accuracy. Minimizes output mean

torchfilter.train.train_dynamics_recurrent(buddy: fannypack.utils.Buddy, dynamics_model: torchfilter.base.DynamicsModel, dataloader: DataLoader, *, loss_function: str = 'nll', log_interval: int = 10) → None

Trains a dynamics model via backpropagation through time.

Parameters:

• buddy (fannypack.utils.Buddy) – Training helper.

• dynamics_model (torchfilter.base.DynamicsModel) – Model to train.

• dataloader (DataLoader) – Loader for a SubsequenceDataset.

Keyword Arguments:

• loss_function (str, optional) – Either “nll” for negative log-likelihood or “mse” for mean-squared error. Defaults to “nll”.

• log_interval (int, optional) – Minibatches between each Tensorboard log.

torchfilter.train.train_dynamics_single_step(buddy: fannypack.utils.Buddy, dynamics_model: torchfilter.base.DynamicsModel, dataloader: DataLoader, *, loss_function: str = 'nll', log_interval: int = 10) → None

Optimizes a dynamics model’s single-step prediction accuracy. This is roughly equivalent to training with train_dynamics_recurrent() with a subsequence length of 2.

Parameters:

• buddy (fannypack.utils.Buddy) – Training helper.

• dynamics_model (torchfilter.base.DynamicsModel) – Model to train.

• dataloader (DataLoader) – Loader for a SingleStepDataset.

Keyword Arguments:

• loss_function (str, optional) – Either “nll” for negative log-likelihood or “mse” for mean-squared error. Defaults to “nll”.

• log_interval (int, optional) – Minibatches between each Tensorboard log.

torchfilter.train.train_filter(buddy: fannypack.utils.Buddy, filter_model: torchfilter.base.Filter, dataloader: DataLoader, initial_covariance: torch.Tensor, *, loss_function: Callable[[torch.Tensor, torch.Tensor], torch.Tensor] = F.mse_loss, log_interval: int = 10, measurement_initialize=False, optimizer_name='train_filter_recurrent') → None

Trains a filter end-to-end via backpropagation through time for 1 epoch over a subsequence dataset.

Parameters:

• buddy (fannypack.utils.Buddy) – Training helper.

• filter_model (torchfilter.base.DynamicsModel) – Model to train.

• dataloader (DataLoader) – Loader for a SubsequenceDataset.

• initial_covariance (torch.Tensor) – Covariance matrix of error in initial posterior, whose mean is sampled from a Gaussian centered at the ground-truth start state. Shape should be (state_dim, state_dim).

Keyword Arguments:

• loss_function (callable, optional) – Loss function, from torch.nn.functional. Defaults to MSE.

• log_interval (int, optional) – Minibatches between each Tensorboard log.

torchfilter.train.train_kalman_filter_measurement(buddy: fannypack.utils.Buddy, measurement_model: torchfilter.base.KalmanFilterMeasurementModel, dataloader: DataLoader, *, loss_function: Callable[[torch.Tensor, torch.Tensor], torch.Tensor] = F.mse_loss, log_interval: int = 10, optimizer_name='train_kalman_filter_measurement') → None

Optimizes a Kalman filter measurement model’s prediction accuracy. Minimizes output mean error only; does not define a loss on uncertainty.

Parameters:

• buddy (fannypack.utils.Buddy) – Training helper.

• measurement_model (torchfilter.base.KalmanFilterMeasurementModel) – Model to train.

• dataloader (DataLoader) – Loader for a SingleStepDataset.

Keyword Arguments:

• loss_function (callable, optional) – Loss function, from torch.nn.functional. Defaults to MSE.

• log_interval (int, optional) – Minibatches between each Tensorboard log.

torchfilter.train.train_particle_filter_measurement(buddy: fannypack.utils.Buddy, measurement_model: torchfilter.base.ParticleFilterMeasurementModel, dataloader: DataLoader, *, loss_function: Callable[[torch.Tensor, torch.Tensor], torch.Tensor] = F.mse_loss, log_interval: int = 10) → None

Reference implementation for pre-training a particle filter measurement model. Minimizes prediction error for log-likelihood outputs from (state, observation) pairs.

Parameters:

• buddy (fannypack.utils.Buddy) – Training helper.

• measurement_model (torchfilter.base.ParticleFilterMeasurementModel) – Model.

• dataloader (DataLoader) – Loader for a ParticleFilterMeasurementDataset.

Keyword Arguments:

• loss_function (callable, optional) – Loss function, from torch.nn.functional. Defaults to MSE.

• log_interval (int, optional) – Minibatches between each Tensorboard log.

torchfilter.train.train_virtual_sensor(buddy: fannypack.utils.Buddy, virtual_sensor_model: torchfilter.base.VirtualSensorModel, dataloader: DataLoader, *, loss_function: Callable[[torch.Tensor, torch.Tensor], torch.Tensor] = F.mse_loss, log_interval: int = 10, optimizer_name='train_measurement') → None

Optimizes a virtual sensor model’s prediction accuracy. Minimizes output mean error only; does not define a loss on uncertainty.

Parameters:

• buddy (fannypack.utils.Buddy) – Training helper.

• measurement_model (torchfilter.base.VirtualSensorModel) – Model to train.

• dataloader (DataLoader) – Loader for a SingleStepDataset.

Keyword Arguments:

• loss_function (callable, optional) – Loss function, from torch.nn.functional. Defaults to MSE.

• log_interval (int, optional) – Minibatches between each Tensorboard log.