[cosc-grad-students-list] Computer Science Master Thesis Defense

[Li, Longzhuang]

Master Thesis Defense
Computer Science Program
Department of Computing Sciences

TITLE:    REAL-TIME HUMAN MOVEMENT PREDICTION BASED ON HUMAN HABITS AND EMOTION
SPEAKER:     Roberto Rodriguez
THESIS CHAIR:     Dr. Longzhuang Li
THESIS COMMITTEE:      Dr. Dulal Kar and Dr. Chen Pan
DATE:    Friday, July 9, 2021
TIME:    10:30 am
PLACE:     WebEx
https://tamucc.webex.com/tamucc/j.php?MTID=m6e121b9df418dd622e464e6faae9e15e<https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Ftamucc.webex.com%2Ftamucc%2Fj.php%3FMTID%3Dm6e121b9df418dd622e464e6faae9e15e&data=04%7C01%7Ccosc-grad-students-list%40listserv.tamucc.edu%7Cc4c92d784fca419df4df08d942290b65%7C34cbfaf167a64781a9ca514eb2550b66%7C0%7C0%7C637613566325480401%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=VWFonaL6eqH5L%2BCVF1C6GKhsf5m2%2FsFyLSxHPuEuo5M%3D&reserved=0>
Meeting number (access code): 177 885 2704
Meeting password: pdVvpGGW457 (73887449 from phones)

ABSTRACT
                                This research project develops a new deep neural network model for real-time human movement prediction based on both human habits and emotion by layering multiple neural networks. By combining multiple residual neural networks (ResNets) using different algorithms on multiple datasets, the new model can increase prediction accuracies and reduce errors because each neural network adjusts to fill in the gaps left behind by the previous neural networks and averages out a proper evaluation of human motion. To be more specific, the new model contains the following components, ResNet34 on Pascal3d+ for motion, FURIA algorithm for emotion, and nearest neighbor neural network for prediction. The result was that the prediction would take 0.05 milliseconds after emotion is selected to start its prediction paths. It does follow the motion at first but with lower accuracy until emotion is given. As its database grew, the accuracy grew as well, leading the prediction to near match real-time movement.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.tamucc.edu/pipermail/cosc-grad-students-list/attachments/20210708/5530b6c1/attachment.htm>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: ATT00001.txt
URL: <http://listserv.tamucc.edu/pipermail/cosc-grad-students-list/attachments/20210708/5530b6c1/attachment.txt>