Title: Experimental study of forecasting of fall using RNN and simulated data.
Presenter: Keshav Bhandari
Advisor: Dr. Anne Ngu
Date/Time: Tuesday, May 4th @12:30 – 1:30 p.m. (CDT)
Zoom Link: https://txstate.zoom.us/j/98587330076
This study presents the lesson learned from the fall-prediction task on stick balancing model. The aim of this study is to exploit mathematical model for data generation, train the model and use it to predict the fall in real human data. Balancing is a complex voluntary motor task that requires the stabilization of a chaotic system, accurately predicting the fall is a significant achievement. The current research focuses on fall detection. However, after a person has fallen, the damage is done in a lot of cases. There is a need to study how we can forecast the onset of falls. We put a fairly simple deep recurrent network into a test and learned important intuitions regarding its success in simulated data and failure in real data.
Deadline: May 5, 2021, midnight