Ilija Bogunovic

alt text 

Postdoctoral Fellow
Learning & Adaptive Systems Group
Department of Computer Science
ETH Zürich

Contact Address:
CAB G 68,
Universitatstrasse 6,
CH-8092 Zürich

E-mail: ilija.bogunovic@inf.ethz.ch


I am a postdoctoral researcher at the ETHZ Learning & Adaptive Systems group led by Prof. Andreas Krause. My research focus is on machine learning algorithms that are theoretically well-founded, robust and practically applicable. I am particularly interested in problems in sequential decision making and topics such as bandits, Bayesian optimization, submodular optimization, and reinforcement learning.

I received my PhD degree from École polytechnique fédérale de Lausanne (EPFL) where I was supervised by Prof. Volkan Cevher and co-supervised by Prof. Jarvis Haupt. Previously, I've done research internships at Twitter and MIT. I received my B.S. degree in electrical engineering from the University of Novi Sad and M.S. degree in computer science from ETH Zurich.

Publications

  1. Overlapping Multi-Bandit Best Arm Identification
    J. Scarlett, I. Bogunovic and V. Cevher
    The 2019 IEEE International Symposium on Information Theory (ISIT), Paris, 2019
    [bibtex] [pdf] [long]

    @inproceedings{scarlett2019multibandit,
    		Author = {Scarlett, Jonathan and Bogunovic, Ilija and Cevher, Volkan},
    		Booktitle = {The 2019 IEEE International Symposium on Information Theory (ISIT)},
    		Title = {Overlapping Multi-Bandit Best Arm Identification},
    		Year = {2019}}	
    	

  2. Robust Adaptive Decision Making: Bayesian Optimization and Beyond
    I. Bogunovic
    EPFL, 2019
    [bibtex] [pdf]

    @article{Bogunovic:262894,
          title = {Robust Adaptive Decision Making: Bayesian Optimization and  Beyond},
          author = {Bogunovic, Ilija},
          publisher = {EPFL},
          address = {Lausanne},
          year = {2019},
          url = {http://infoscience.epfl.ch/record/262894},
          doi = {10.5075/epfl-thesis-9147}}
    	

  3. Adversarially Robust Optimization with Gaussian Processes
    I. Bogunovic, J. Scarlett, S. Jegelka and V. Cevher
    (spotlight) Conference on Neural Information Processing Systems (NeurIPS), Montreal, 2018
    [bibtex] [pdf] [slides]

    @inproceedings{bogunovic2018adversarially,
    		Author = {Bogunovic, Ilija and Scarlett, Jonathan and Jegelka, Stefanie and Cevher, Volkan},
    		Booktitle = {Conference on Neural Information Processing Systems (NeurIPS)},
    		Title = {Adversarially Robust Optimization with Gaussian Processes},
    		Year = {2018}}
    	

  4. Robust Maximization of Non-Submodular Objectives
    I. Bogunovic*, J. Zhao* and V. Cevher
    International Conference on Artificial Intelligence and Statistics (AISTATS), Lanzarote, Canary Islands, 2018
    [bibtex] [pdf]

    @inproceedings{bogunovic2018robust,
    	Author = {Bogunovic, Ilija and Zhao, Junyao and Cevher, Volkan},
    	Booktitle = {International Conference on Artificial Intelligence and Statistics (AISTATS)},
    	Title = {Robust Maximization of Non-Submodular Objectives},
    	Year = {2018}}
    

  5. High Dimensional Bayesian Optimization via Additive Models with Overlapping Groups
    P. T. Y. Rolland, J. Scarlett, I. Bogunovic and V. Cevher
    International Conference on Artificial Intelligence and Statistics (AISTATS), Lanzarote, Canary Islands, 2018
    [bibtex] [pdf]

    @inproceedings{rolland2018high,
    	Author={Rolland, Paul and Scarlett, Jonathan and Bogunovic, Ilija and Cevher, Volkan},
    	Booktitle = {International Conference on Artificial Intelligence and Statistics (AISTATS)},
    	Title = {High-Dimensional Bayesian Optimization via Additive Models with Overlapping Groups},
    	Year = {2018}}
    

  6. Streaming Robust Submodular Maximization: A Partitioned Thresholding Approach
    S. Mitrovic, I. Bogunovic, A. Norouzi Fard, J. Tarnawski and V. Cevher
    Conference on Neural Information Processing Systems (NIPS), Long Beach, 2017
    [bibtex] [pdf]

    @inproceedings{mitrovic2017streaming,
      title={Streaming Robust Submodular Maximization: A Partitioned Thresholding Approach},
      author={Mitrovi{\'c}, Slobodan and Bogunovic, Ilija and Norouzi-Fard, Ashkan and Tarnawski, Jakub M and Cevher, Volkan},
      booktitle={Advances in Neural Information Processing Systems (NIPS)},
      pages={4557--4566},
      year={2017}}
    

  7. A Distributed Algorithm for Partitioned Robust Submodular Maximization
    I. Bogunovic, S. Mitrovic, J. Scarlett and V. Cevher
    International Workshop on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP), Curacao, 2017
    [bibtex] [pdf]

    @inproceedings{bogunovic2017distributed,
      title={A Distributed Algorithm for Partitioned Robust Submodular Maximization},
      author={Bogunovic, Ilija and Mitrovi{\'c}, Slobodan and Scarlett, Jonathan and Cevher, Volkan},
      booktitle={Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP), 2017 IEEE 7th International Workshop on},
      pages={1--5},
      year={2017},
      organization={IEEE}
    }
    

  8. Robust Submodular Maximization: A Non-Uniform Partitioning Approach
    I. Bogunovic, S. Mitrovic, J. Scarlett and V. Cevher
    International Conference on Machine Learning (ICML), Sydney, 2017
    [bibtex] [pdf]

    @inproceedings{bogunovic2017robust,
    	Author = {Bogunovic, Ilija and Mitrovi{\'c}, Slobodan and Scarlett, Jonathan and Cevher, Volkan},
    	Booktitle = {International Conference on Machine Learning (ICML)},
    	Title = {Robust Submodular Maximization: A Non-Uniform Partitioning Approach},
    	Year = {2017}}
    

  9. Lower Bounds on Regret for Noisy Gaussian Process Bandit Optimization
    J. Scarlett, I. Bogunovic and V. Cevher
    Conference on Learning Theory (COLT), Amsterdam, 2017
    [bibtex] [pdf]

    @inproceedings{scarlett2017lower,
    	Author = {Scarlett, Jonathan and Bogunovic, Ilija and Cevher, Volkan},
    	Booktitle = {International Conference on Learning Theory (COLT)},
    	Title = {Lower Bounds on Regret for Noisy Gaussian Process Bandit Optimization},
    	Year = {2017}}
    

  10. An Efficient Streaming Algorithm for the Submodular Cover Problem
    A. Norouzi Fard, A. Bazzi, M. El Halabi, I. Bogunovic, Y.-P. Hsieh and V. Cevher
    Conference on Neural Information Processing Systems (NIPS), Barcelona, 2016
    [bibtex] [pdf]

    @inproceedings{norouzi2016efficient,
      title={An efficient streaming algorithm for the submodular cover problem},
      author={Norouzi-Fard, Ashkan and Bazzi, Abbas and Bogunovic, Ilija and El Halabi, Marwa and Hsieh, Ya-Ping and Cevher, Volkan},
      booktitle={Advances in Neural Information Processing Systems (NIPS)},
      pages={4493--4501},
      year={2016}
    }
    

  11. Truncated Variance Reduction: A Unified Approach to Bayesian Optimization and Level-Set Estimation
    I. Bogunovic, J. Scarlett, A. Krause and V. Cevher
    Conference on Neural Information Processing Systems (NIPS), Barcelona, 2016
    [bibtex] [pdf]

    @inproceedings{bogunovic2016truncated,
      title={Truncated variance reduction: A unified approach to Bayesian optimization and level-set estimation},
      author={Bogunovic, Ilija and Scarlett, Jonathan and Krause, Andreas and Cevher, Volkan},
      booktitle={Advances in Neural Information Processing Systems (NIPS)},
      pages={1507--1515},
      year={2016}
    }
    

  12. Time-Varying Gaussian Process Bandit Optimization
    I. Bogunovic, J. Scarlett and V. Cevher
    International Conference on Artificial Intelligence and Statistics (AISTATS), Cadiz, 2016
    [bibtex] [pdf]

    @inproceedings{bogunovic2016time,
      title={Time-varying Gaussian process bandit optimization},
      author={Bogunovic, Ilija and Scarlett, Jonathan and Cevher, Volkan},
      booktitle={Artificial Intelligence and Statistics (AISTATS)},
      pages={314--323},
      year={2016}
    }
    

  13. Learning-Based Compressive Subsampling
    L. Baldassarre, Y.-H. Li, J. Scarlett, B. Gözcü, I. Bogunovic and V. Cevher
    IEEE Journal on Selected Topics in Signal Processing
    [bibtex] [pdf]

    @article{baldassarre2016learning,
      title={Learning-based compressive subsampling},
      author={Baldassarre, Luca and Li, Yen-Huan and Scarlett, Jonathan and G{\"o}zc{\"u}, Baran and Bogunovic, Ilija and Cevher, Volkan},
      journal={IEEE Journal of Selected Topics in Signal Processing},
      volume={10},
      number={4},
      pages={809--822},
      year={2016},
      publisher={IEEE}
    }
    

  14. Active Learning of Self-concordant like Multi-index Functions
    I. Bogunovic, V. Cevher, J. Haupt and J. Scarlett
    International Conference on Acoustics, Speech and Signal Processing (ICASSP), Brisbane, 2015
    [bibtex] [pdf]

    @inproceedings{bogunovic2015active,
      title={Active learning of self-concordant like multi-index functions},
      author={Bogunovic, Ilija and Cevher, Volkan and Haupt, Jarvis and Scarlett, Jonathan},
      booktitle={Acoustics, Speech and Signal Processing (ICASSP), 2015 IEEE International Conference on},
      pages={2189--2193},
      year={2015},
      organization={IEEE}
    }
    

  15. Near-Optimally Teaching the Crowd to Classify
    A. Singla, I. Bogunovic, G. Bartok, A. Karbasi and A. Krause
    International Conference on Machine Learning (ICML), Beijing, 2014
    [bibtex] [pdf]

    @inproceedings{singla2014near,
      title={Near-Optimally Teaching the Crowd to Classify.},
      author={Singla, Adish and Bogunovic, Ilija and Bart{\'o}k, G{\'a}bor and Karbasi, Amin and Krause, Andreas},
      booktitle={ICML},
      pages={154--162},
      year={2014}
    }