Team Name: ServeML
Team Members:
- Prudhviraj Naidu (B.S Computer Science) - naidupr@mail.uc.edu
Faculty Advisor: Dr. Nan Niu - niunn@ucmail.uc.edu
Project Topic Area: Machine Learning and Deep Learning: What Happens After Training?
Deep learning has started to dominate traditional Machine Learning problems. In this project, we investigate if deep learning systems can truly adapt to unseen changes in serving data. We look at common neural networks used for image classification. These neural networks are trained on a standard training set. We apply numerous image distortions to the test set to see how models’ accuracies change.
There isn't enough research of analyzing ML solutions as a software tools. In terms of how they behave to common case erroneous situations, or how they behave with respect to one another.
I have worked over a year as a Computer Vision/ML Engineer. Therefore, I am interested as to how these ML solutions work as software tools. In terms, could we measure their error rates with serving data? When is the optimial time to update them?
There is only one member on the team. I work for 2 hours on Friday, Saturday and Sundays. I email my advisor with any progress and schedule meetings with him to discuss future steps.