A signal-processing-based technique for P300 evoked potential detection with the applications into automated character recognition

EURASIP Journal on Advances in Signal Processing

Table 3 Listings of the P300 detection results on the eight-word testing dataset

Words for testing	M = 15	M = 14	M = 13	M = 12	M = 10	M = 8	M = 5	M = 3
FOOD	FOOD	FOOD	FOOD	FN OD	FB OD	FNC D	FON D	FON D
MOOT	MOOT	MOOT	MOOT	MOOT	MOOT	MOOT	MOOT	MOOZ
HAM	HAM	HAM	HAM	HAM	HAM	HAM	B AA	BCE
PIE	PIE	PIE	PIE	PIE	PIE	PIE	PIE	PIE
CAKE	CAKE	CAKE	CAKE	CAKE	CAKE	CAKE	CAKE	CALC
TUNA	TUNA	TUNA	TO NA	TO NA	TO NA	TO NA	TOB A	TC NJ
ZYGOT	ZYGOT	ZYA OT	ZYGOT	Z5A OT	ZSA OT	ZSA OT	ZSA ON	ZSA ON
4567	4567	4567	4567	4567	4567	4567	458 7	R 58 7
Accuracy%	100.00	96.77	96.77	87.10	87.10	83.87	70.97	54.84

This dataset is for a variety of M values (results on experiment III). Note that the P300 template employed here in this experiment was obtained from a ten-word training dataset, which was different from the one we used in experiments I and II. Similarly, each wrong detection character is set in italics.