Adaptive Deep Supervised Autoencoder Based Image Reconstruction for Face Recognition

Table 3

(a) Training time (seconds) for different methods. (b) Testing time (seconds) for different methods. The proposed method costs the least amount of testing time comparing with other methods.

(a)

Methods	Time
DAE [12]	8.61
MDAE [17]	10.54
CAE [15]	7.86
DLN [31]	23.43
SSAE [9]	53.51
RICA [32]	13.44
TDRM [20]	110.2
Our method	122.32

(b)

Methods	Time
DAE [12]	0.27
MDAE [17]	0.3
CAE [15]	0.26
DLN [31]	0.35
SSAE [9]	0.22
RICA [32]	0.19
TDRM [20]	0.18
Our method	0.13