Sequence to Sequence encoder decoder framework on points data coming from videos.

Predicting conflict intensity during human-human interactions using audiovisual data from real-world videos. I am utilizing a bi-directional LSTM as an encoder and a 2 layer stacked LSTM coupled with attention mechanism for my decoder.