qumphy.models.minirocket module

File: qumphy/models/minrocket.py Project: 22HLT01 QUMPHY Contact: oskar.pfeffer@ptb.de Gitlab: https://gitlab.com/qumphy Description: MiniRocKeT implementation as from […].

class qumphy.models.minirocket.MiniRocket(*args: Any, **kwargs: Any)[source]

Bases: Module

MiniRocket model with feature extractor and prediction head.

forward(x)[source]

Run a forward pass through the MiniRocket model.

Parameters:: x (torch.Tensor) – Input tensor of shape (batch_size, c_in, seq_len).
Returns:: Model output after feature extraction, prediction head, and output activation.
Return type:: torch.Tensor

class qumphy.models.minirocket.MiniRocketFeatures(*args: Any, **kwargs: Any)[source]

Bases: Module

This is a Pytorch implementation of MiniRocket developed by Malcolm McLean and Ignacio Oguiza This module extracts MiniRocket features from time-series data using fixed convolutional kernels, multiple dilations, and proportion of positive values features.

MiniRocket paper citation: @article{dempster_etal_2020,

author = {Dempster, Angus and Schmidt, Daniel F and Webb, Geoffrey I}, title = {{MINIROCKET}: A Very Fast (Almost) Deterministic Transform for Time Series Classification}, year = {2020}, journal = {arXiv:2012.08791}

} Original paper: https://arxiv.org/abs/2012.08791 Original code: https://github.com/angus924/minirocket

extract_features(data)[source]

Extract MiniRocket features from input data.

Parameters:: data (torch.Tensor) – Input data of shape (batch_size, seq_len) or (batch_size, c_in, seq_len).
Returns:: Extracted MiniRocket features.
Return type:: torch.Tensor

fitting = False

forward(x)[source]

Extract MiniRocket features from an input tensor.

Parameters:: x (torch.Tensor) – Input tensor of shape (batch_size, c_in, seq_len).
Returns:: Extracted MiniRocket features of shape (batch_size, num_features).
Return type:: torch.Tensor

get_quantiles(num_quantiles)[source]

Calculate quantile values using the golden ratio.

Parameters:: num_quantiles (int) – Number of quantile values to calculate.
Returns:: List containing the calculated quantile values.
Return type:: list

kernel_size = 9

num_kernels = 84

class qumphy.models.minirocket.MiniRocketHead(*args: Any, **kwargs: Any)[source]: Bases: Sequential

qumphy.models.minirocket.get_minirocket_features(o, model, chunksize=1024, use_cuda=None, to_np=True)[source]

Extract MiniRocket features from a large dataset in chunks.

Parameters:

o (np.ndarray or torch.Tensor) – Input dataset.
model (nn.Module) – MiniRocket feature extraction model.
chunksize (int) – Number of samples processed in each chunk.
use_cuda (bool) – If True, use CUDA. If False, use CPU. If None, CUDA is used when available.
to_np (bool) – If True, return the features as a NumPy array. If False, return them as a torch.Tensor.

Returns:

Extracted MiniRocket features.

Return type:

np.ndarray or torch.Tensor