Robust Deep Network with Maximum Correntropy Criterion for Seizure Detection

<table>Model analysis of two important parameters of R-SAE. (a) Seizure detection performance under different feature numbers; (b) seizure detection performance with different selections of <svg height="8.3000002" id="M149" style="vertical-align:-0.13794pt" version="1.1" viewbox="0 0 9.7250004 8.3000002" width="9.7250004" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink">
<g transform="matrix(.017,-0,0,-.017,.062,8.038)"><path d="M548 455q-32 -77 -74 -77q-37 0 -103 10l-3 -2q36 -30 48.5 -62t12.5 -80q0 -103 -75 -179.5t-175 -76.5q-70 0 -113 45t-43 128q0 115 83.5 200.5t209.5 85.5q31 0 77.5 -4t67.5 -4q36 0 62 30zM350 274q0 54 -17 86q-19 35 -57 35q-50 0 -90 -37.5t-59 -91.5t-19 -109
q0 -61 24.5 -96.5t65.5 -35.5q51 0 87.5 46.5t50.5 100.5t14 102z" id="x1D70E"></path></g>
</svg>. In this figure, SEN-FT and SPE-FT are sensitivity and specificity after fine-tuning and SEN-NFT and SPE-NFT are those before fine-tuning.</table>

BioMed Research International

fig6

Figure 6

Figure 6: Robust Deep Network with Maximum Correntropy Criterion for Seizure Detection 