top of page
Deep Learning Based Dynamic Patterns Classification of Videos Using Spatio-Temporal Virtual Communication
S. P. Kasthuri Arachchi, Yi-Cheng Chen, Timothy K. Shih
National Central University, Taiwan (R. O. C.)
Dual-CNNVCLSTM model with learning rate schedules
(a)
(c)
(b)
(d)
Performance of Dual-CNNVCLSTM using different learning rate schedules (a) with constant LR; (b) with step decay schedule; (c) with time-based decay schedule; (d) with exponential decay schedule.
Dual-CNNVCLSTM model with adaptive learning rates
(a)
(b)
(c)
Performance of Dual-CNNVCLSTM using different adaptive learning rate methods (a) with AdaGrad optimization; (b) with RMSProp optimization; (c) Adam optimization.
bottom of page