BCI2000 Wiki - Recent changes [en]

User Reference:Timing

Mellinger — Mon, 29 Jun 2026 13:01:48 GMT

Timing Display

← Older revision		Revision as of 13:01, 29 June 2026
Line 33:		Line 33:
	*real-time operation -- roundtrip time should stay below theoretical sample block length;		*real-time operation -- roundtrip time should stay below theoretical sample block length;
	*module communication overhead -- this is represented by the difference between ''roundtrip'' and ''stimulus'' curves;		*module communication overhead -- this is represented by the difference between ''roundtrip'' and ''stimulus'' curves;
	*stimulus trigger time and stimulus display -- each yellow dot represents a triggered stimulus, each green dot represents a stimulus display;		*stimulus trigger time and stimulus display -- each yellow dot represents a triggered stimulus, each green dot represents a visual stimulus display;
	*audio playback -- each violet dot represents the transition from a zero audio buffer to a nonzero one, relative to a preceding yellow dot (stimulus trigger).		*audio playback -- each violet dot represents the transition from a zero audio buffer to a nonzero one, relative to a preceding yellow dot (stimulus trigger).

File:BCI2000Analysis AnalysisWindow.png

Mellinger — Tue, 23 Jun 2026 17:29:33 GMT

Mellinger uploaded a new version of File:BCI2000Analysis AnalysisWindow.png

New page

File:BCI2000Analysis AnalysisWindow.png

Mellinger — Tue, 23 Jun 2026 17:22:16 GMT

Mellinger uploaded a new version of File:BCI2000Analysis AnalysisWindow.png

New page

User Reference:BCI2000Analysis

Mellinger — Tue, 23 Jun 2026 17:12:01 GMT

Viewing data as seen by the Application

← Older revision		Revision as of 17:12, 23 June 2026
Line 87:		Line 87:
	If you want to apply the CSP Spatial Filter to the chain without saving it to a file, choose ''File→Apply Spatial Filter Parameters''.		If you want to apply the CSP Spatial Filter to the chain without saving it to a file, choose ''File→Apply Spatial Filter Parameters''.

	==Viewing data as seen by the Application==		==Viewing data as seen by the Application Module==
	Finally, after training and applying classifier and/or spatial filter, you may be interested in the distribution of data as seen by the Application module. To this end, select the Application node as a target (its actual name depends on the kind of data analyzed).		Finally, after training and applying classifier and/or spatial filter, you may be interested in the distribution of data as seen by the Application module. To this end, select the Application node as a target (its actual name depends on the kind of data analyzed).

User Reference:P300Classifier

Mellinger — Tue, 23 Jun 2026 13:34:56 GMT

Classifier Weight Normalization

← Older revision		Revision as of 13:34, 23 June 2026
(2 intermediate revisions by the same user not shown)
Line 255:		Line 255:

	==Classifier Weight Normalization==		==Classifier Weight Normalization==
	When applying classifier weights to obtain a classification score, and ~~select~~ a target with the largest score, this procedure is invariant to multiplication of all classifier weights with a constant factor. We are, thus, free to normalize classifier weights with a factor that allows us to interpret classification scores in a certain way.		When applying classifier weights to obtain a classification score, and when selecting a target with the largest score, this procedure is invariant against multiplication of all classifier weights with a constant factor.
	As described [[Programming_Reference:StimulusTask_Class#Accumulation_and_Combination_of_Evidence\|here]], a normalization that makes within-class variance identical to unity ~~when the classifier is applied to the training data set~~ will result in classification scores to represent an estimate of the log evidence in favor of the respective target. This allows the AccumulateEvidence and MinimumEvidence parameters to work.		We are, thus, free to normalize classifier weights with a factor that allows us to interpret classification scores in a certain way.

			As described [[Programming_Reference:StimulusTask_Class#Accumulation_and_Combination_of_Evidence\|here]], a normalization that makes within-class variance identical to unity will result in classification scores to represent an estimate of the log evidence in favor of the respective target. This allows the [[Programming_Reference:StimulusTask_Class#Accumulation_and_Combination_of_Evidence\|AccumulateEvidence and MinimumEvidence parameters]] to work.

	==Future work==		==Future work==

	* Present classification results as topographies and graphs as the Matlab-based P300 GUI (has been done in BCI2000Analysis)		* Present classification results as topographies and graphs as the Matlab-based P300 GUI (has been done in [[User_Reference:BCI2000Analysis\|BCI2000Analysis]])

	* Let the P300Classifier GUI work with frequency-based features (has been done in BCI2000Analysis)		* Let the P300Classifier GUI work with frequency-based features (has been done in [[User_Reference:BCI2000Analysis\|BCI2000Analysis]])

	* Speed up the P300Classifier GUI by using multithreading when solving the Least Squares (QR factorization)		* Speed up the P300Classifier GUI by using multithreading when solving the Least Squares (QR factorization)