finished (2020/02/20, finished 2020/10/18)| SUTD was established in 2009 as a collaboration with MIT, and has recently in 2018 topped the list of emerging engineering schools in the world. At SUTD, there is a strong emphasis on interdisciplinary and collaborative research. Our unconventional naming of departments (or pillars) is a reflection of our philosophy that impactful research would naturally encompass multiple disciplines. SUTD is located in Singapore, a multiracial multicultural country, where English is the main language used in all schools and most workplaces, especially among people of different races. We are seeking two highly motivated research fellows with strong algebraic backgrounds, who are interested to explore new research directions in deep learning, and more generally in Artificial Intelligence. A successful candidate can expect to conduct research on a completely new topic that combines commutative algebra with AI, and will work closely with other team members. The goal is to develop an algebraic framework for understanding the architecture hyperparameters of deep neural networks, especially in relation to unsupervised feature learning. This framework shall be based on methods from both commutative algebra and computational algebra. We shall also build upon this algebraic framework and develop algorithms for generating optimal neural network architectures that are able to perform optimized feature learning on unlabeled data sets. Duration of position: 2 years. (Two post-doctoral positions available.)
|
)