Greg Hamerly is an Associate Professor of Computer Science at Baylor University, where his research interests lie in the areas of supervised and unsupervised machine learning and the applications of these areas to other fields.
He has worked recently on developing a machine learning approach to identifying so-called "white-eye" in recreational photography, which is an indication of many different types of eye disease, including pediatric retinoblastoma. The result of this work is a mobile application called CRADLE which can be used to screen for symptoms of eye disease. The app has been downloaded over 100,000 times around worldwide, and has been instrumental in early detection of disease for a number of people.
His background includes a research program called SimPoint, which applies unsupervised learning models (i.e. vector-space clustering algorithms) to the task of reducing time in detailed computer architecture simulations. With SimPoint, a joint collaboration with researchers at UCSD and Intel, computer architecture designers can test the performance characteristics of new CPU designs in hours, rather than months, on standard benchmark software.
Dr. Hamerly is also interested in improving both the automation and efficiency of clustering algorithms such as k-means. His work has enabled the ability to automatically determine the number of clusters in a vector-based dataset (using the G-means and PG-means algorithms). He has also developed a very simple k-means acceleration that is at least twice as efficient as other algorithms, even other accelerated algorithms, especially in low-dimensional data.