Ozge Akmandor is a PhD candidate at Princeton University, advised by Prof. Niraj K. Jha. She received the BS degree in electrical and electronics engineering from Middle East Technical University, Turkey, in 2015, and the MA degree in electrical engineering from Princeton University, NJ, in 2017. Her research interests lie at the intersection of machine learning, natural language processing, and Internet-of-Things, with a focus on smart healthcare applications. She is the recipient of six Dr. Altay Awards and Princeton University PhD fellowship. Her research was featured in a spotlight article in IEEE Computer Magazine and IEEE Transactions on Multiscale Computing Systems. Two patent applications have been filed based on her research.
Research Abstract: Semantically Enhanced Classification of Real-world Tasks
Traditional approaches use supervised machine learning (ML) algorithms to distill intelligence from data and, hence, impart smartness to various devices. They try to map the training data to the corresponding labels while ensuring generalizability to unseen data. However, they do not take advantage of meaning-based relationships among labels in the decision process. Consider a dataset that has a calm sleep, REM sleep and stress situation as labels. Current supervised ML algorithms will result in the same data-label model even if we replace the labels with class 1,class 2, and class 3. However, calm sleep and REM sleep are semantically more similar, but less similar to stress situation. It would be advantageous to exploit this semantic relationship during classification. Our framework, called SECRET, addresses the above problem through a dual-space classification approach.