Table 8.

10-fold cross validation accuracies (in %) for the classification of VALENCE using a multichannel DNN on the DEAP dataset. $F_i$ refers to fold number i. The best performance for each fold is highlighted in bold.

Approach	F1	F2	F3	F4	F5	F6	F7	F8	F9	F10	Average
Bi-modal AE [7]	-	-	-	-	-	-	-	-	-	-	85.20
TTO	87.67	87.03	87.85	86.93	87.26	87.44	88.03	87.08	87.75	87.25	87.43
VAE-transfer	85.17	84.86	83.92	85.48	84.75	84.06	84.23	85.42	84.90	84.69	84.75
CNN-transfer	90.89	91.12	90.22	90.39	90.51	90.27	90.71	90.39	91.08	90.84	90.64