Computer Sciences Commons^™

Overcoming Roadblocks In Introducing Virtual World Technology To High Schools, Casey Dylan Bailey

Graduate Theses and Dissertations

The EAST (Environmental And Spatial Technology) Initiative is a non-profit educational organization that provides students in over two hundred schools in eight states with access to advanced computing technologies for the purpose of enabling students to develop technical skills early and to produce solutions to local community problems. Although many high-end technologies are available through EAST, they are desktop solutions that individual students use and there are none that enable students within a school or between schools to collaborate.

This thesis is a saga that documents the identification and removal of many roadblocks to introducing a 3D multi-user virtual simulation …

A Comparison Of Dropout And Weight Decay For Regularizing Deep Neural Networks, Thomas Grant Slatton

Computer Science and Computer Engineering Undergraduate Honors Theses

In recent years, deep neural networks have become the state-of-the art in many machine learning domains. Despite many advances, these networks are still extremely prone to overfit. In neural networks, a main cause of overfit is coadaptation of neurons which allows noise in the data to be interpreted as meaningful features. Dropout is a technique to mitigate coadaptation of neurons, and thus stymie overfit. In this paper, we present data that suggests dropout is not always universally applicable. In particular, we show that dropout is useful when the ratio of network complexity to training data is very high, otherwise traditional …