Jamell Dozier

Jamell  Dozier
Undergraduate Research and Innovation Scholar
Advisor: Tomaso Poggio
Department: EECS
Years: 2017-2018
Research Project Title:

Learning Insideness with Deep Neural Networks

abstract:*The "inside/outside" problem remains a challenging problem in machine learning due to the property that a small change in the input can alter the solution. Through testing various neural network models and comparing their performances both with each other as well as with the performance of networks evaluating the mnist dataset we hope to gain further insight into the extent to which neural networks can be relied on to tackle similar classification problems.
About:

"I view Super UROP both as a chance to further enhance my experience with research and an opportunity to continue my exploration into machine learning. I have had some prior experience using neural networks, and I hope to emerge from this project with an even better understanding of their behavior. Above all, I am excited for the opportunity to study the lesser-known aspects of a neural network's capabilities and shortcomings."