minerva.transforms.tfc

Classes

TFC_Transforms

Transformations used in the TFC model.

Module Contents

class minerva.transforms.tfc.TFC_Transforms

Bases: minerva.transforms.transform._Transform

Transformations used in the TFC model. It consists of time and frequency domain data augmentation.

DataTransform_FD(sample)

Weak and strong augmentations. Consists of jittering and adding or removing frequency components.

Parameters

• sample: np.ndarray

  The input data to be augmented

Returns

• np.ndarray

  The augmented data

Parameters:

sample (numpy.ndarray)

Return type:

numpy.ndarray

DataTransform_TD(sample, jitter_ratio=0.8)

Weak and strong augmentations. Consists of jittering and removing time components.

Parameters

• sample: np.ndarray

  The input data to be augmented

• jitter_ratio: float

  The ratio of the jittering transformation

Returns

• np.ndarray

  The augmented data

Parameters:

• sample (numpy.ndarray)

• jitter_ratio (float)

Return type:

numpy.ndarray

__call__(x)

Method that applies the transformations to the input data.

Parameters

• x: Union[np.ndarray, torch.Tensor]

  The input data to be transformed

Returns

• Tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]

  A tuple with the original data, the transformed data in the time domain the frequency version of the data and the tranformed data in frequency domain

Parameters:

x (Union[numpy.ndarray, torch.Tensor])

Return type:

Tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]

add_frequency(x, pertub_ratio=0)

function to add frequency components to the input data.

Parameters

• x: np.ndarray

  The input data to be augmented

• pertub_ratio: float

  The ratio of the frequency components to be added

Returns

• np.ndarray

  The data with added frequency components

Parameters:

• x (numpy.ndarray)

• pertub_ratio (float)

jitter(x, sigma=0.8)

Add noise to the input data.

Parameters

• x: np.ndarray

  The input data to be augmented

• sigma: float

  The standard deviation of the noise

Returns

• np.ndarray

  The data with added noise

Parameters:

• x (numpy.ndarray)

• sigma (float)

one_hot_encoding(X, n_values=None)

One-hot encoding of the input data

Parameters

• X: np.ndarray

  The input data to be encoded

• n_values: int

  The number of classes in the data. If None, the number of classes is inferred from the data

Returns

• np.ndarray

  The one-hot encoded data

Parameters:

• X (numpy.ndarray)

• n_values (int)

remove_frequency(x, maskout_ratio=0)

function to remove frequency components from the input data.

Parameters

• x: np.ndarray

  The input data to be augmented

• maskout_ratio: float

  The ratio of the frequency components to be removed

Returns

• np.ndarray

  The data with removed frequency components

Parameters:

• x (numpy.ndarray)

• maskout_ratio (float)