Undergrad Research Project - Predicting Human Behavior Using Deep Learning

Spring 2015

Benjamin Siegel
Radu Marculescu
Project description

The goal of this project is to use the deep learning model to be able to analyze and eventually predict human behavior. This will be done through several steps. The first step is to build the input data so that we can define the information that we are looking to analyze. We will then do calculations on this data using a Recurrent Convolutional Markov Model. Next, we will use these results to interpret different filters from the data. These filters will then be matched up with different with segments of video recordings so that they can correspond to specific actions. The next step is to try to use these results to make a label track that is able to predict human actions. We will compare performance of our model with other human behavior prediction models. Finally, we will use our data to try to account for anomalies in human behavior from occurring in our predictions. Through all of this we plan to be able to use this method to be able to better recognize and predict human behavior.

Return to project list