Roi Livni

I am a research instructor at the Department of Computer Science at Princeton University. My research focuses on Theoretical Machine Learning. I received my Ph.D from the Center for Brain Sciences (ELSC) at The Hebrew University of Jerusaelm under the supervision of Amir Globerson.

Contact
Princeton University
Department of Computer Science
35 Olden St
Princeton NJ 08540

Email: rlivni at cs.princeton.edu
Room No: 219


Teaching


Spring 2017: Theoretical Machine Learning (COS 511)


Publications

Effective Semisupervised Learning on Manifolds
A. Globerson, R. Livni, S. Shalev-Shwartz
(To appear at COLT, 2017)
[pdf]

Bandits with Movement Costs and Adaptive Pricing
T. Koren, R. Livni, Y. Mansour
(To appear at COLT, 2017)
[pdf]

Learning Infinite--Layer Networks: Without the Kernel Trick
R. Livni, D. Carmon, A. Globerson
(To appear at ICML, 2017)
[pdf]

Online Pricing With Strategic and Patient Buyers,
M. Feldman, T. Koren, R. Livni, Y. Mansour, A. Zohar.
Advances of Neural Information and Processing Systems 29 (NIPS), 2016
[pdf]

Online Learning With Low Rank Experts,
E. Hazan, T. Koren, R. Livni, Y. Mansour
29th Conference on Learning Theory (COLT), 2016
[pdf]

Improper Deep Kernels,
U.Heinemann, R. Livni, E. Eban, G. Elidan, A. Globerson.
19th International Conference on Artificial Intelligence and Statistics (AISTAT), 2016
[pdf],

Classification with Low Rank and Missing Data,
E. Hazan, R. Livni, Y. Mansour
32nd International Conference on Machine Learning (ICML), 2015
[pdf], [code]

On the Computational Efficiency of Training Neural Networks,
R. Livni, S. Shalev-Shwartz and O. Shamir
Advances in Neural Information Processing Systems 27 (NIPS), 2014
[pdf],

Honest Compressions and Their Application to Compression Schemes,
R. Livni and P. Simon
26th Conference on Learning Theory (COLT), 2013. (Mark Fulk best student paper award)
[pdf]

Vanishing Component Analysis,
R.Livni, D. Lehavi, S. Schein, H. Nachlieli, S Shalev-Shwartz and A. Globerson
30th International Conference on Machine Learning (ICML),2013. (Best paper award)
[pdf ] [code]

A Simple Geometric Interpretation of SVM using Stochastic Adversaries,
R. Livni, K. Crammer and A. Globerson
15th International Conference on Artificial Intelligence and Statistics (AISTAT), 2012.
[pdf], [supplementary]

On Extreme Points of the Dual Ball of a Polyhedral Space,
R. Livni
Extracta Mathematicae.24(3): 219-241, 2009.
[pdf]


Preprints/Tech Reports


An Algorithm for Training Polynomial Networks
R. Livni, S.Shalev-Shwartz and O. Shamir,
[pdf]