Skip to main content
. Author manuscript; available in PMC: 2018 Jan 15.
Published in final edited form as: Neuroimage. 2016 Apr 11;145(Pt B):314–328. doi: 10.1016/j.neuroimage.2016.04.003

Figure 2.

Figure 2

Schematics of the DBN pretraining followed by the DNN fine-tuning exemplified for three hidden layers. The input layer consisted of 74,484 in-brain voxels of an fMRI volume, each hidden layer consisted of 100 hidden nodes, and the output layer consisted of four nodes to assign each of the four tasks (i.e., LH, RH, AD, and VS). The DBN and DNN were trained using 1,320 fMRI volumes of input samples across 11 training subjects (i.e., 30 volumes per task for each subject), and the trained DNN was validated using 120 fMRI volumes from one remaining validation subject. LH, left-hand clenching; RH, right-hand clenching; AD, auditory attention; VS, visual stimulus.