Publications:

Machine Learning, Statistics and Applications


  1. Bullet“The EM Algorithm is Sample-Optimal for Learning Mixtures of Well-Separated Gaussians,” with Jeongyeol Kwon.

  2. To Appear in the Proceedings of the Conference on Learning Theory (COLT), 2020.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“Disentangling Mixtures of Epidemics on Graphs,” with Jessica Hoffmann, Soumya Basu and Surbhi Goel.

  2. To Appear in the Proceedings of the International Conference on Machine Learning (ICML), 2020.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“High Dimensional Robust M-Estimation: Arbitrary Corruption and Heavy Tails,” with Liu Liu and Tianyang Li.

  2. Submitted, 2019.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“Mix and Match: An Optimistic Tree-Search Approach for Learning Models from Mixture Distributions,” with Karthikeyan Shanmugam, Rajat Sen, Matthew Faw and Sanjay Shakkottai.

  2. Submitted, 2020.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“Communication-Efficient Asynchronous Stochastic Frank-Wolfe over Nuclear Norm Balls” with Qi Lei, Jiacheng Zhuo and Alex Dimakis.

  2. To Appear in the Proceedings of AISTATS, 2020.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“High Dimensional Robust Sparse Regression,” with Liu Liu, Yanyao Shen and Tianyang Li.

  2. To Appear in the Proceedings of AISTATS, 2020.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“EM Converges for a Mixture of Many Linear Regressions,” with Jeongyeol Kwon.

  2. To Appear in the Proceedings of AISTATS, 2020.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“Approximate Newton-based Statistical Inference using only stochastic Gradients,” with Tianyang Li, Anastasios Kyrillidis and Liu Liu.

  2. Submitted, 2020.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“Primal-Dual Block Frank-Wolfe” with Qi Lei, Jiacheng Zhuo, Alex Dimakis and Inderjit Dhillon.

  2. To Appear in the proceedings of Neural Information Processing Systems (NeurIPS), 2019.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“Low Rank Matrix Regression under Heavy Tailed Distributions,” with Liu Liu and Tianyang Li.

  2. Submitted, 2019.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“Robust Gaussian Graphical Model Estimation of Tree Structured Graphs,” with Ashish Katiyar.

  2. Submitted, 2019.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“Entropic Latent Variable Discovery,” with Murat Kocaoglu, Sanjay Shakkottai, Alex Dimakis, and Sriram Vishwanath.

  2. Submitted, 2019

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  5. Bullet“Learning Graphs from Noisy Epidemic Cascades,” with Jessica Hoffmann.

  6. To appear in the Proceedings of SIGMETRICS, 2019.

  7. Avalailable: Abstract. Paper PDF. ArXiv PDF.

  8. Second Prize in the INFORMS Nicholson Competition


  1. Bullet“Global Convergence of the EM Algorithm for Mixtures of Two Component Linear Regression,” with Jeongyeol Kwon, Wei Qian, Yudong Chen and Damek Davis. 2019.

  2. To appear in the Proceedings of the Conference on Learning Theory (COLT), 2019.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“Robust Estimation of Tree Structured Gaussian Graphical Models,” with Ashish Katiyar and Jessica Hoffmann.

  2. To appear in the Proceedings of the International Conference on Machine Learning (ICML), 2019.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“Finding Low-rank Solutions to Matrix Problems, Efficiently and Provably,” with Dohyung Park, Anastasios Kyrillidis and Sujay Sanghavi.

  2. SIAM J. Imaging Science, Vol. 11, No. 4, pp. 2165-2876, 2018.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“The Cost of Uncertainty in Curing Epidemics,” with Jessica Hoffmann.

  2. Proceedings of the ACM Sigmetrics Conference, 2018.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“Statistical Inference Using SGD,” with Tianyang Li, Liu Liu and Anastasios Kyrillidis.

  2. Proceedings of AAAI, 2018.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“Exploiting Latent Attack Semantics for Intelligent Malware Detection,” with Mikhail Kazdagli, Sanjay Shakkottai and Mohit Tiwari, 2017.

  2. Avalailable: Abstract. Paper PDF.


  1. Bullet“Fast Algorithms for Robust PCA via Gradient Descent,” with Xinyang Yi, Dohyung Park and Yudong Chen.

  2. Advances in Neural Processing Systems (NeurIPS), 2016.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“More Supervision, Less Computation: Statistical-Computational Tradeoffs in Weakly Supervised Learning” with Xinyang Yi, Zhaoran Wang, Zhuoran Yang and Han Liu.

  2. Advances in Neural Processing Systems (NeurIPS), 2016.

  3. Avalailable: Abstract. Paper PDF.


  1. Bullet“Solving a Mixture of Many Random Linear Equations by Tensor Decomposition and Alternating Minimization” with Xinyang Yi and Sujay Sanghavi. 2016.

  2. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“Statistical Optimization in High Dimensions,” with Huan Xu and Shie Mannor.

  2. Operations Research,Vol. 64, No. 4, pp. 958-979, 2016.

  3. Avalailable: Abstract. Paper PDF.


Partial preliminary results appeared in The Proceedings of AISTATS, 2012 (available here: Paper PDF)


  1. Bullet“Matrix Completion with Column Manipulation: Near Optimal Sample-Robustness-Rank Tradeoffs,” with Y. Chen, H. Xu and S. Sanghavi.

  2. IEEE Transactions on Information Theory, Vol. 62, No. 1, pp. 503-526, 2015.

  3. Avalailable: Abstract. Paper PDF.


Partial preliminary results appeared at the International Conference on Machine Learning (ICML), 2011.


  1. Bullet“Distinguishing Infections on Different Graph Topologies,” with C. Milling, S. Mannor and S. Shakkottai.

  2. IEEE Transactions on Information Theory, Vol. 61, No. 6, pp. 3100-3120, 2015.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


Partial preliminary results appeared in The Proceedings of SIGMETRICS, 2012, under the title “Network Forensics: Random Infection vs. Spreading Epidemic” (available here: Paper PDF).


A different subset of the results appeared in The Proceedings of The Allerton Conference on Communication, Control and Computing, 2012, under the title “On Identifying the Causative Network of an Epidemic” (available here: Paper PDF).


  1. Bullet“Regularized EM Algorithms: A Unified Framework and Statistical Guarantees,” with Xinyang Yi.

  2. To Appear in Advances in Neural Processing Systems (NeurIPS), 2015.

  3. Avalailable: Abstract. Paper PDF.


  1. Bullet“Optimal Linear Estimation under Unknown Nonlinear Transform,” with Xinyang Yi, Zhaoran Wang and Han Liu.

  2. To Appear in Advances in Neural Processing Systems (NeurIPS), 2015.

  3. Avalailable: Abstract.


  1. Bullet“Binary Embedding: Fundamental Limits and a Fast Algorithm,” with Xinyang Yi and Eric Price.

  2. Proceedings of the International Conference on Machine Learning (ICML), 2015.

  3. Avalailable: Abstract. Paper PDF.


  1. Bullet“Detecting Cascades from Weak Signatures,” with Eli Meirom, Shie Mannor, Ariel Orda and Sanjay Shakkottai.

  2. Submitted, 2015.

  3. Avalailable: Abstract.


  1. Bullet“FrogWild! -- Fast PageRank Approximations on Graph Engines,” with Ioannis Mitliagkas, Michael Borokhovich and Alex Dimakis.

  2. To Appear in the Proceedings of the 41st International Conference on Very Large Data Bases (VLDB), 2015.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“Localized Epidemic Detection in Networks with Overwhelming Noise,” with Eli Meirom, Chris Milling, Shie Mannor, Ariel Orda and Sanjay Shakkottai.

  2. To Appear as a short paper in the Proceedings of the ACM SIGMETRICS Conference, 2015.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“Local Detection of Infections in Heterogeneous Networks,” with Chris Milling, Shie Mannor and Sanjay Shakkottai.

  2. To appear in the Proceedings of INFOCOM, 2015.

  3. Avalailable: Abstract. Paper PDF. ArXiv PDF.


  1. Bullet“Greedy Subspace Clustering,” with Dohyung Park and Sujay Sanghavi.

  2. To appear in the Proceedings of Neural Information Processing Systems (NeurIPS), 2014.

  3. Avalailable: Abstract. Paper PDF. Project page.


  1. Bullet“A Convex Formulation for Mixed Regression: Minimax Optimal Rates,” with Yudong Chen and Xinyang Yi.

  2. The Proceedings of the Conference on Learning Theory (COLT), 2014.

  3. Avalailable: Abstract. Paper PDF. Full Paper PDF.


Partial preliminary results appear under the title “A Convex Formulation for Mixed Regression: Near Optimal Rates in the Face of Noise” (available here: ArXiv PDF).


  1. Bullet“Alternating Minimization for Mixed Linear Regression,” with Xinyang Yi and Sujay Sanghavi.

  2. To Appear in the Proceedings of the International Conference on Machine Learning (ICML), 2014.

  3. Avalailable: Abstract. ArXiv PDF.


  1. Bullet“Finding Dense Subgraphs Through Low-Rank Approximations,” with Dimitris Papailiopoulos, Ioannis Mitliagkas and Alex Dimakis.

  2. To Appear in the Proceedings of the International Conference on Machine Learning (ICML), 2014.

  3. Avalailable: Abstract. Paper PDF.


  1. Bullet“Modeling the Time-Varying Subjective Quality of HTTP Video Streams with Rate Adaptations,” with Chao Chen, Lark Kwon Choi, Gustavo de Veciana, Robert Heath, Jr., and Al C. Bovik.

  2. IEEE Transactions on Image Processing, Vol. 23, No. 5, pp. 2206-2221, 2014.

  3. Avalailable: Abstract. Paper PDF.


  1. Bullet“Efficient Algorithms for Budget-Constrained Markov Decision Processes,” with Ned Dimitrov and David P. Morton.

  2. IEEE Transactions on Automatic Control, Vol. 59, No. 10, pp. 2813-2817, 2014.

  3. Avalailable: Abstract. Paper PDF.


  1. Bullet“Streaming PCA with Many Missing Entries,” with Ioannis Mitliagkas and Prateek Jain.

  2. Submitted, 2014.

  3. Avalailable: Abstract. Paper PDF.


  1. Bullet“Detecting Epidemics Using Highly Noisy Data,” with C. Milling, S. Mannor and S. Shakkottai.

  2. The Proceedings of the ACM Int. Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc), 2013.

  3. Avalailable: Abstract. Paper PDF.


  1. Bullet“Streaming, Memory-Limited Principal Component Analysis,” with I. Mitliagkas and P. Jain.

  2. To appear in The Proceedings of the Neural Information Processing Systems (NeurIPS), 2013.

  3. Avalailable: Abstract. Paper PDF.


  1. Bullet“Robust High Dimensional Sparse Regression and Matching Pursuit,” with Y. Chen and S. Mannor.

  2. The Proceedings of the International Conference on Machine Learning (ICML), 2013.

  3. ArXiv:1301.2725, 2013.

  4. Avalailable: Abstract. ICML PDF. ArXiv PDF.


  1. Bullet“Noisy and Missing Data Regression: Distribution-Oblivious Support Recovery,” with Y. Chen.

  2. The Proceedings of the International Conference on Machine Learning (ICML), 2013.

  3. Avalailable: Abstract. ICML PDF.


  1. Bullet“Outlier-Robust PCA: The High Dimensional Case,” with H. Xu and S. Mannor

  2. IEEE Transactions on Information Theory, Vol. 59, No. 1, pp. 546-572, 2013.

  3. Avalailable: Abstract. Paper PDF.


Partial preliminary results appeared at The Allerton Conference on Communication, Control and Computing, and The International Conference on Learning Theory (COLT). COLT Paper PDF


  1. Bullet“Low-rank Matrix Recovery from Errors and Erasures,” with Y. Chen, A. Jalali and S. Sanghavi.

  2. IEEE Transactions on Information Theory,  Vol. 59, No. 7, pp. 4324-4337, 2013.

  3. Avalailable: Abstract. Paper PDF.


Partial preliminary results appeared at the International Symposium on Information Theory (ISIT), 2011.


  1. Bullet“Optimization Under Probabilistic Envelope Constraints,” with H. Xu and S. Mannor.

  2. Operations Research, Vol. 60, No. 3, pp. 682-699, 2012.

  3. Avalailable: Abstract. Paper PDF.


  1. Bullet“A Distributional Interpretation of Robust Optimization,” with S. Mannor and H. Xu.

  2. Mathematics of Operations Research, Vol. 37, No. 1, pp. 95-110, 2012.

  3. Avalailable: Abstract. Paper PDF.


Partial preliminary results appeared at The Allerton Conference on Communication, Control and Computing, 2010.


  1. Bullet“User Rankings from Comparisons: Learning Permutations in High Dimensions,” with I. Mitliagkas, A. Gopalan and S. Vishwanath.

  2. Proceedings of The Allerton Conference on Communications, Control and Computing, 2011.

  3. Avalailable: Abstract.


  1. Bullet“Sparse Algorithms are not Stable: a No-Free-Lunch Theorem,” with H. Xu and S. Mannor

  2. IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 34, No. 1, pp. 187-193, 2012.

  3. Avalailable: Abstract. Paper PDF.


Partial preliminary results appeared at The Allerton Conference on Communication, Control and Computing, 2008.


  1. Bullet“Robust PCA via Outlier Pursuit,” with H. Xu and S. Sanghavi.

  2. IEEE Transactions on Information Theory, Vol 58, No. 5, pp. 3047-3064, 2012.

  3. Avalailable: Abstract. Paper PDF.


Partial preliminary results appeared in the proceedings of Neural Information Processing Systems (NeurIPS). 2010


  1. Bullet“Robust Optimization in Machine Learning,” with H. Xu and S. Mannor

  2. Book Chapter in Optimization for Machine Learning, S. Sra, S. Nowozin, S. Wright., Editors, MIT Press.

  3. To Appear in 2011.

  4. Avalailable: Abstract. Paper PDF.


  1. Bullet“Exploiting Sparse Dynamics for Bandwidth Reduction in Cooperative Sensing Systems,” with H. Ganapathy and L. Ying.

  2. IEEE Transactions on Signal Processing, Vol. 61, No. 14, pp. 3671-3682, 2013.

  3. Available: Abstract. Paper PDF. ArXiv PDF.


Partial preliminary results appeared under the title “Limited Feedback for Cognitive Radio Networks Using Compressed Sensing,” at the Allerton Conference on Communication, Control and Computing, 2010.


  1. Bullet“Reinforcement Learning for Link Adaptation in MIMO-OFDM,” with S. Yun

  2. Proceedings of Globecom. 2010

  3. Avalailable: Abstract. Paper PDF.


  1. Bullet“Robust Regression and Lasso,” with H. Xu and S. Mannor

  2. IEEE Transaction on Information Theory, Vol. 56, No. 7, pp. 3561-3574. 2010.

  3. Avalailable: Abstract. Paper PDF.


Partial preliminary results appeared in the Proceedings of the Neural Information Processing Systems Conference (NeurIPS), December 2008


  1. Bullet“Adaptation in Convolutionally-Coded MIMO-OFDM Wireless Systems through Supervised Learning and Subcarrier Ordering,” with R.C. Daniels, and R.W. Heath, Jr.

    IEEE Transactions on Vehicular Technology, Vol. 59, No. 1, pp. 114-126. 2010.

  1. Avalailable: Abstract. Paper PDF.


Partial preliminary results appeared under the title “A Supervised Learning Approach to Adaptation in Practical MIMO-OFDM Wireless Systems,” in the Proceedings of Globecom 2008


  1. Bullet“Multiclass Support Vector Machines for Adaptation in MIMO-OFDM Wireless Systems,” with S. Yun.

    The Proceedings of the Allerton Conference on Communication, Control, and Computing, September 2009

  1. Avalailable: Abstract. Paper PDF.


  1. Bullet“Risk Sensitive Robust Support Vector Machines,” with S. Mannor, H. Xu, S. Yun.

    In The Proceedings of the Conference on Decision and Control (CDC) December 2009.

  1. Avalailable: Abstract. Paper PDF.


  1. Bullet“Rank Minimization via Online Learning,” with I. Dhillon, P. Jain, and R. Meka.

  2. In The Proceedings of the International Conference on Machine Learning (ICML), 2008.

  3. Avalailable: Abstract. Paper PDF.


  1. Bullet“Learning in the Limit with Adversarial Disturbances,” with S. Mannor.

  2. In The Proceedings of the International Conference on Learning Theory (COLT), 2008.

  3. Avalailable: Abstract. Paper PDF.


  1. Bullet“Robustness and Regularization of Support Vector Machines,” with H. Xu and S. Mannor.

    Journal of Machine Learning Research (JMLR),  Vol. 10, pp. 1485-1510. 2009.

  1. Avalailable: Abstract. Paper PDF.


  1. Bullet“A Bayesian Approach to Data Driven Optimization Under Uncertainty,” with S. Mannor.

    The Proceedings of the Allerton Conference on Communication, Control, and Computing, September 2007.


  1. Bullet“Adaptability via Sampling,” with D. Bertsimas.

    In The Proceedings of the Conference on Decision and Control (CDC) December 2007.


  1. Bullet“An Inequality for Nearly Log-concave Distributions with Applications to Learning,” with S. Mannor

    IEEE Transactions on Information Theory, Vol. 53, No.3, pp. 1043-1057, 2007

  1. Avalailable: Abstract. Paper PDF.


Partial preliminary results appeared in the Proceedings of The International Conference on Learning Theory (COLT), 2004.


Machine Learning, Statistics and Applications. I am interested in a variety of problems at the intersection of statistics, machine learning and optimization. One key theme is the interconnection between robustness and structure (like sparsity, or low-rank). Another theme is understanding “robust” statistics -- problems where data are corrupted, noisy or missing, models mis-specified, etc.  At the application level, I am interested in developing data-driven algorithms for a variety of applications, mostly related to various aspects of wireless networks. Please e-mail me for conference versions, if they are not available. [The abstract function does not yet work... but it will soon....]



Papers in chronological order, appearing once once

Papers organized along other central themes: (Papers may appear in multiple categories)

  1. BulletOptimization and Applications

  2. BulletWireless/Networks

  3. BulletOther Engineering Applications