KULProcessedDataset¶

class torcheeg.datasets.KULProcessedDataset(root_path: str = './preprocessed_data', chunk_size: int = 64, overlap: int = 0, num_channel: int = 64, online_transform: None | Callable = None, offline_transform: None | Callable = None, label_transform: None | Callable = None, before_trial: None | Callable = None, after_trial: None | Callable = None, after_subject: None | Callable = None, io_path: None | str = None, io_size: int = 1048576, io_mode: str = 'lmdb', num_worker: int = 0, verbose: bool = True)[source][source]¶

This dataset contains EEG recordings from 16 subjects listening to one of two competing speech audio streams, after matlab preprocessing. This class generates training samples and test samples according to the given parameters, and caches the generated results in a unified input and output format (IO). The relevant information of the dataset is as follows:

Author: Biesmans et al.
Year: 2016
Download URL: https://zenodo.org/records/4004271
Reference: Biesmans W, Das N, Francart T, et al. Auditory-inspired speech envelope extraction methods for improved EEG-based auditory attention detection in a cocktail party scenario[J]. IEEE transactions on neural systems and rehabilitation engineering, 2016, 25(5): 402-412.
Stimulus: Four Dutch short stories, lasting approximately six minutes each. Either dry speech was offered, i.e., each speaker was presented to a different ear, or speech signals were processed by (dead room) head-related transfer functions, simulating a more realistic listening scenario in which the speakers are spatially located 90 degrees to the left and the right of the subject.
Signals: The BioSemi ActiveTwo system was used to record 64-channel EEG signals at 8196 Hz sample rate (downsampled to 64 Hz). Each subject completed 8 trials, each lasting 6 minutes.
Labels: Attended direction (attended_ear=’L’ for left, attended_ear=’R’ for right), stimulus condition (condition=’dry’ or ‘hrtf’).

In order to use this dataset, the download file 4004271.zip is required. After unzipping, the folder should contain the following files:

4004271/
├── preprocess_data.m
├── S1.mat
├── S2.mat
├── S3.mat
├── ...
└── S16.mat

After running preprocess_data.m, preprocessed_data folder should be generated and contain the following files:

preprocessed_data/
├── S1.mat
├── S2.mat
├── S3.mat
├── ...
└── S16.mat

An example dataset for CNN-based methods:

from torcheeg.datasets import KULProcessedDataset
from torcheeg import transforms

dataset = KULProcessedDataset(root_path='./preprocessed_data',
                           offline_transform=transforms.To2d(),
                           online_transform=transforms.ToTensor(),
                           label_transform=transforms.Compose([
                               transforms.Select('attended_ear'),
                               transforms.Lambda(lambd=lambda x: 1 if x == 'L' else 0)
                           ]))
print(dataset[0])
# EEG signal (torch.Tensor[1, 64, 64]),
# label (int)

Parameters:

root_path (str) – Path to the preprocessed data files in MATLAB format. (default: './preprocessed_data')
chunk_size (int) – Number of data points included in each EEG chunk as training or test samples. If set to -1, the EEG signal of an entire trial is used as a single chunk. (default: 64)
overlap (int) – The number of overlapping data points between different chunks when dividing EEG data. (default: 0)
num_channel (int) – Number of channels to use. The dataset contains 64 channels. (default: 64)
online_transform (Callable, optional) – The transformation applied to the EEG signals. The input is a np.ndarray, and the output is used as the first value of each element in the dataset. (default: None)
offline_transform (Callable, optional) – The usage is the same as online_transform, but executed before generating IO intermediate results. (default: None)
label_transform (Callable, optional) – The transformation applied to the label. The input is an information dictionary, and the output is used as the second value of each element in the dataset. (default: None)
before_trial (Callable, optional) – A hook function performed on the trial to which the sample belongs. It is executed before the offline transformation and is typically used to implement context-dependent sample transformations, such as moving averages. The input of this hook function is a 2D EEG signal with shape (number of electrodes, number of data points), and the ideal output shape is also (number of electrodes, number of data points).
after_trial (Callable, optional) – A hook function performed on the trial to which the sample belongs. It is executed after the offline transformation and is typically used to implement context-dependent sample transformations, such as moving averages. The input and output of this hook function should be a sequence of dictionaries representing a sequence of EEG samples. Each dictionary contains two key-value pairs, indexed by eeg (the EEG signal matrix) and key (the index in the database) respectively.
after_subject (Callable, optional) – A hook function performed on the subject to which the sample belongs. It is executed after the offline transformation and is typically used to implement context-dependent sample transformations, such as moving averages. The input and output of this hook function should be a sequence of dictionaries representing a sequence of EEG samples. Each dictionary contains two key-value pairs, indexed by eeg (the EEG signal matrix) and key (the index in the database) respectively.
io_path (str) – The path to the generated unified data IO, cached as an intermediate result. If set to None, a random path will be generated. (default: None)
io_size (int) – Maximum size the database may grow to; used to size the memory mapping. If the database grows larger than map_size, an exception will be raised and the user must close and reopen. (default: 1048576)
io_mode (str) – Storage mode of EEG signals. When io_mode is set to lmdb, TorchEEG provides an efficient database (LMDB) for storing EEG signals. LMDB may not perform well on limited operating systems, where a file system-based EEG signal storage is also provided. When io_mode is set to pickle, pickle-based persistence files are used. When io_mode is set to memory, memory is used. (default: lmdb)
num_worker (int) – Number of subprocesses to use for data loading. 0 means that the data will be loaded in the main process. (default: 0)
verbose (bool) – Whether to display logs during processing, such as progress bars. (default: True)

KULProcessedDataset¶

Docs

Tutorials

Resources