Supervised and unsupervised spatial-filter decompositions for M/EEG decoding. More...

Classes
class	DecodingCsp
	Common Spatial Patterns decoder for two-class discriminative spatial filtering. More...
struct	IcaLabelResult
	Outcome of labelling a single ICA component. More...
class	MlIcaLabel
	Static utility that labels ICA components against EOG/ECG references and a muscle spectral heuristic. More...
class	DecodingSpoc
	Source Power Comodulation decoder for continuous-target regression on band-power. More...
class	DecodingSsd
	Spatio-spectral decomposition for narrowband signal enhancement on continuous M/EEG. More...

Enumerations
enum class	IcaComponentLabel { Brain , Eog , Ecg , Muscle , Other }
	Categorical label assigned to a single ICA component. More...

Functions
DECODINGSHARED_EXPORT const char *	buildDateTime ()
DECODINGSHARED_EXPORT const char *	buildHash ()
DECODINGSHARED_EXPORT const char *	buildHashLong ()

Detailed Description

Supervised and unsupervised spatial-filter decompositions for M/EEG decoding.

DECODINGLIB exposes a small, deliberately scikit-learn-shaped API (fit, transform, fitTransform, inverseTransform) around the four spatial-filter families that dominate BCI and biomarker work on continuous and epoched M/EEG: CSP for binary class-discriminative power, SPoC for regressing a continuous target onto band-power, SSD for noise-aware narrowband enhancement, and an ICA component labeller for automatic artefact tagging. The public surface mirrors mne.decoding so that a pipeline prototyped in MNE-Python can be ported one-to-one into a real-time mne-cpp application.

Every algorithm reduces to a generalised eigendecomposition of two symmetric positive-definite covariance matrices, computed inline with Eigen rather than delegated to LAPACK; this keeps the library free of heavyweight numerical dependencies and makes it trivially usable from the WebAssembly build. MNE-specific extensions on top of the upstream Python API are the explicit TransformMode enums (AveragePower vs raw CspSpace projection), opt-in log or z-score normalisation of band-power features, and a closed-form inverse-transform back to sensor space.

Enumeration Type Documentation

◆ IcaComponentLabel

enum class DECODINGLIB::IcaComponentLabel

strong

Categorical label assigned to a single ICA component.

Spans the four artefact families that account for essentially all non-brain variance in scalp M/EEG — ocular, cardiac and muscular — plus a generic Other bucket for components that fail every specific check (line noise, channel pops, residual sensor jumps). Brain is the default and means "keep this component in the reconstruction".

Enumerator
Brain
Eog
Ecg
Muscle
Other

Definition at line 77 of file decoding_ica_label.h.

Function Documentation

◆ buildDateTime()

const char * DECODINGLIB::buildDateTime ( )

Returns build date and time.

Definition at line 29 of file decoding_global.cpp.

◆ buildHash()

const char * DECODINGLIB::buildHash ( )

Returns abbreviated build git hash.

Definition at line 33 of file decoding_global.cpp.

◆ buildHashLong()

const char * DECODINGLIB::buildHashLong ( )

Returns full build git hash.

Definition at line 37 of file decoding_global.cpp.

Classes

Enumerations