Skip to main content

Advertisement

Table 3 SVM Model 1 Coefficients Extracted by Stage of Working Memory

From: Machine learning identification of EEG features predicting working memory performance in schizophrenia and healthy adults

WM Stage Frequency Feature Weight
Baseline   Accuracy = .77
   Central theta 2 −1.127
   Occipital beta −0.183
   Frontal theta 1 −0.158
   Occipital theta 2 −0.143
  intercept 0.480
Encode   Accuracy = .96
   Frontal gamma −1.693
   Central gamma −0.748
  intercept 0.607
Retain   Accuracy = .77
   Central theta 1 −1.192
   Occipital theta 2 −0.307
  intercept 0.500
Retrieve   Accuracy = .88
   Occipital gamma −1.380
   Occipital theta 2 −0.646
   Central gamma −0.261
  intercept 0.597
  1. Four separate SVM Models were constructed in HN with entry of 60 EEG Features by WM Stage. Features extracted by 1-norm SVM to classify correct vs. incorrect trials with correct trials labeled 1 and incorrect labeled −1. Intercepts of four models were equivalent. All other features weighted at 0