Exception to be triggered when DN3-configuration parsing fails.

Maps various channel sets into the Deep10-10 scheme, and normalizes data between [-1, 1] with an additional scaling parameter to describe the

Implements complete convolution block with order: - Convolution - dropout (spatial) - activation - batch-norm - (op

Exception to be triggered when DN3-dataset variants load NaN data, or data becomes NaN when pushed through transforms.

Collects thinkers, each of which may collect multiple recording sessions of the same tasks, into a dataset with (largely) consistent: -

This objects contains non-critical meta-data that might need to be tracked for :py:`Dataset` objects. Generally not necessary to be construct

Parses DN3 configuration files. Checking the DN3 token for listed datasets.

Interface for bridging mne Raw instances as PyTorch compatible "Dataset". Parameters ---------- raw : mne.io.Raw Raw data,

Collects multiple recordings of the same person, intended to be of the same task, at different times or conditions.

A session-wise implementation of He & Wu 2019; https://doi.org/10.1109/TBME.2019.2913914 Used to some success with DNNs in Kostas & Rudzicz 2

Z-score normalization of trials

T-fixup assumes self-attention norms are removed

Exception to be triggered when DN3-dataset-specific issues arise.

Parses dataset entries in DN3 config

this loss performs label smoothing to compute cross-entropy with soft labels, when smoothing=0.0, this is the same as torch.nn.CrossEntropyLoss

Initialization of the Base Trainable object. Any learning procedure that leverages DN3atasets should subclass this base class. By defaul

A more wav2vec 2.0 style of constrastive self-supervision, more inspired-by than exactly like it.

A generic Classifer container. This container breaks operations up into feature extraction and feature classification to enable convenience i

This is a base model used by the provided models in the library that is meant to make those included in this library as powerful and multi-pu

This is the DN3 re-implementation of Lawhern et. al.'s EEGNet from: https://iopscience.iop.org/article/10.1088/1741-2552/aace8c Notes

This is the DN3 re-implementation of Lawhern et. al.'s EEGNet from: https://iopscience.iop.org/article/10.1088/1741-2552/aace8c Notes

Scale the input to range from low to high

In effect, simply an implementation of linear kernel (multi)logistic regression

Base class for various preprocessing actions. Sub-classes are called with a subclass of `_Recording` and operate on these instances in-place.

The Thinker Invariant Densenet from Kostas & Rudzicz 2020, https://doi.org/10.1088/1741-2552/abb7a7 This alone is not strictly "thinker inva

Abstract base class for any supported recording