yjustc2019 / cnn-convlstm-time-series Goto Github PK
View Code? Open in Web Editor NEWThis project forked from gauravanand25/cnn-convlstm-time-series
Inspired by the success and computational efficiency of convolutional architectures for various sequential tasks compared to recurrent neural networks. We explored CNN and RCNN autoencoder whose representations can be utilized for the task of time-series classification. Our results surpass existing RNN and DTW-based-classifiers on 11 out of 30 datasets while the existing RNN achieved 8/30.