Toward Measuring Target Perception: First-Order and Second-Order Deep Network Pipeline for Classification of Fixation-Related Potentials

Figure 10

t-SNE of 25% raw data and outputs from each layer of our model. (a) Raw data; (b) outputs from CONV-Net; (c) outputs from ACOVP layer; (d) outputs from SPDC-Net. Red and blue dots represent the target and nontarget trials, respectively.

(a)

(b)

(c)

(d)