Junier Bárbaro Oliva

Hello

I'm Junier Oliva, an assistant professor in the computer science department at UNC Chapel Hill. I am the head of the LUPA Lab, a research lab devoted to performing machine learning and artificial intelligence. Currently I'm looking to bridge the gap between human and machine learning by extending our paradigm past single point estimation and considering collections like distributions, sets, sequences, and graphs. By extending approaches to a more holistic view, we develop methods that not only make assessments on entire collections at a time, but also make use of related collections of points to provide context when considering a particular instance.

Moreover, I'm interested in exporting concepts from learning on distributional and functional inputs to modern techniques in deep learning. I have a preference for simple estimators that make few assumptions; in particular I'm interested in (frequentist) non-parametric methods. One very interesting challenge is scaling up non-parametric methods to huge datasets and high dimensions. This work will help us solve problems like predicting whether a Twitter trending topic will go viral, or predicting the risk of disease given a person's functional brain data, or predicting the future distribution of dark matter particles. See my research statement, and LUPA Lab projects page for more info!

Prior to completing my Ph.D. in Machine Learning at Carnegie Mellon University, I also received my B.S. and M.S. in Computer Science from Carnegie Mellon University. Also, I spent a year as a software engineer for Yahoo!, and a summer as a machine learning intern at Uber ATG.

Research Interests

Machine learning, artificial intelligence, nonparametric statistics, deep learning, statistical data mining, signal processing, graphical models, generative models, kernel methods, scalability, complex datasets, optimization, density estimation.

Papers

J. Oliva, A. Dubey, M. Zaheer, B.Póczos, R. Salakhutdinov, E. Xing, and J. Schneider. Transformation Autoregressive Networks. International Conference on Machine Learning (ICML). Stockholm, Sweden, 2018.

J. Oliva, B. Póczos, and J. Schneider. The Statistical Recurrent Unit. International Conference on Machine Learning (ICML). Sydney, Australia, 2017.

S. Ravanbakhsh, J. Oliva, S. Fromenteau, L. Price, S. Ho, J. Schneider, & B. Póczos. Estimating cosmological parameters from the dark matter distribution. International Conference on Machine Learning (ICML). NYC, NY, 2016.

K. Kandasamy, G. Dasarathy, J. Oliva, J. Schneider, & B. Póczos. Gaussian process bandit optimisation with multi-fidelity evaluations. Advances in Neural Information Processing Systems (NIPS). Barcelona, Spain, 2016.

X. Wang, J. Oliva, J. Schneider, B. Póczos. Nonparametric Risk and Stability Analysis for Multi-Task Learning Problems. The International Joint Conference on Artificial Intelligence (IJCAI). NYC, NY, 2016.

J. Oliva*, A. Dubey*, A. Wilson, B. Póczos, E. Xing, J. Schneider. Bayesian Nonparametric Kernel-Learning. International Conference on Artificial Intelligence and Statistics (AISTATS). Cadiz, Spain, 2016.

D. Sutherland*, J. Oliva*, B. Póczos, J. Schneider. Linear-time Learning on Distributions with Approximate Kernel Embeddings. Association for the Advancement of Artificial Intelligence (AAAI). Phoenix, AZ, 2016.

J. Oliva, W. Neiswanger, B. Póczos, E. Xing, and J. Schneider. Fast Function to Function Regression. International Conference on Artificial Intelligence and Statistics (AISTATS). San Diego, CA, 2015.

J. Oliva, B. Póczos, T. Verstynen, A. Singh, J. Schneider, F.C. Yeh, and E.Y. Tseng. FuSSO: Functional Shrinkage and Selection Operator. International Conference on Artificial Intelligence and Statistics (AISTATS). Reykjavik, Iceland, 2014.

J. Oliva, W. Neiswanger, B. Póczos, J. Schneider, and E. Xing. Fast Distribution to Real Regression. International Conference on Artificial Intelligence and Statistics (AISTATS). Reykjavik, Iceland, 2014.

J. Oliva, B. Póczos, and J. Schneider. Distribution to Distribution Regression. International Conference on Machine Learning (ICML). Atlanta, GA, 2013.