Photo of Igor Burago

Igor Burago

2014 – 2019
Research Interests:
  • Adaptive machine learning in wireless systems
  • Adversarial machine learning
  • Automatic problem generation
Education:
  • 2014
    M.S. in Computer and Information Science
    University of Oregon, U.S.
  • 2011
    M.S. in Applied Mathematics and Informatics
    Far Eastern Federal University, Russia

Selected Experience

2017
Software engineering summer intern at Google, Mountain View, California. Developed a prototype of an experimental semi-supervised entropy-based filter of spam videos and suspicious activity using YouTube session data.
2015–2016
Teaching assistant for the Intermediate Programming (in Python) and Data Structure Implementation and Analysis (in C++) courses at the Department of Computer Science, University of California, Irvine. Rated 7.9–8.8/9.0 over 5 quarters.
2012–2014
Network and telecom services engineering graduate research fellow at the Information Services Department, University of Oregon. Participated in prototyping of network systems built on the principles of software-defined networking.

Publications

No matches

Book Chapter

  • Aakanksha Chowdhery
    Marco Levorato
    Igor Burago
    Sabur Baidya
    Amir M. Rahmani
    Pasi Liljeberg
    Jürgo-Sören Preden
    Axel Jantsch
    A. Chowdhery, M. Levorato, I. Burago, and S. Baidya, Urban IoT Edge Analytics, in Fog Computing in the Internet of Things, A. M. Rahmani, P. Liljeberg, J.-S. Preden, and A. Jantsch, Eds. Springer, 2018, ch. 6, pp. 101 – 120. Release
    @incollection{chowdhery-levorato-2018-fogiot,
      author = {Aakanksha Chowdhery and Marco Levorato and Igor Burago and Sabur Baidya},
      title = {Urban {IoT} Edge Analytics},
      editor = {Amir M. Rahmani and Pasi Liljeberg and J{\"u}rgo-S{\"o}ren Preden and Axel Jantsch},
      booktitle = {Fog Computing in the Internet of Things},
      chapter = {6},
      pages = {101--120},
      publisher = {Springer},
      year = {2018},
      doi = {10.1007/978-3-319-57639-8_6},
    }

Journal Articles

  • Korosh Vatanparvar
    Sina Faezi
    Igor Burago
    Marco Levorato
    Mohammad Abdullah Al Faruque
    K. Vatanparvar, S. Faezi, I. Burago, M. Levorato, and M. A. Al Faruque, Extended Range Electric Vehicle With Driving Behavior Estimation in Energy Management, IEEE Transactions on Smart Grid, vol. 10, no. 3, pp. 2959 – 2968, May 2019. Release

    Battery and energy management methodologies have been proposed to address the design challenges of driving range and battery lifetime in Electric Vehicles (EV). However, the driving behavior is a major factor which has been neglected in these methodologies. In this paper, we propose a novel context-aware methodology to estimate the driving behavior in terms of future vehicle speeds and integrate this capability into EV energy management. We implement a driving behavior model using a variation of Artificial Neural Networks (ANN) called Nonlinear AutoRegressive model with eXogenous Inputs (NARX). We train our novel context-aware NARX model based on historical behavior of real drivers, their recent driving reactions, and route average speed retrieved from Google Maps in order to enable driver-specific and self-adaptive driving behavior modeling and long-term estimation. We analyze the estimation error of our methodology and its impact on a battery lifetime-aware automotive climate control, comparing to the state-of-the-art methodologies for various estimation window sizes. Our methodology shows only 12% error for up to 30-second speed prediction which is an improvement of 27% compared to the state-of-the-art. Therefore, the higher accuracy helps the controller to achieve up to 82% of the maximum energy saving and battery lifetime improvement achievable in ideal methodology where the future vehicle speeds are known.

    @article{vatanparvar-faezi-2019-tsg,
      author = {Korosh Vatanparvar and Sina Faezi and Igor Burago and Marco Levorato and Mohammad Abdullah {Al~Faruque}},
      title = {Extended Range Electric Vehicle With Driving Behavior Estimation in Energy Management},
      journal = {IEEE Transactions on Smart Grid},
      volume = {10},
      number = {3},
      pages = {2959--2968},
      month = may,
      year = {2019},
      doi = {10.1109/tsg.2018.2815689},
    }
  • Igor Burago
    Marco Levorato
    I. Burago and M. Levorato, Network Estimation in Cognition-Empowered Wireless Networks, IEEE Transactions on Cognitive Communications and Networking, vol. 1, no. 2, pp. 244 – 256, Jun. 2015. Release

    An approach to parametric identification of the transmission processes of the terminals in a wireless network is proposed, presenting a trade-off between accuracy of capturing the temporal dependencies in observations of transmission processes and the time complexity of the estimation procedure. The maximum likelihood estimator is built for an approximation of the true likelihood function for the observed network activity. A complex network where terminals store packets in a finite buffer and implement a backoff-based random channel access protocol is considered. Minimal information is available for observation to the cognitive terminals, in the form of energy readings mapped to the number of transmitting nodes in each time instant. The entanglement of the transmission processes induced by interference and the filtering effect of packet buffering make this task particularly difficult. It is shown how, based on the estimated parameters, the cognitive terminals, operating in the same channel resource, can predict the transmission trajectories of the other nodes and devise smart transmission strategies controlling the interference generated to the network.

    @article{burago-levorato-2015-tccn,
      author = {Igor Burago and Marco Levorato},
      title = {Network Estimation in Cognition-Empowered Wireless Networks},
      journal = {IEEE Transactions on Cognitive Communications and Networking},
      volume = {1},
      number = {2},
      pages = {244--256},
      month = jun,
      year = {2015},
      doi = {10.1109/tccn.2016.2517013},
    }

Conference Proceedings

  • Igor Burago
    Marco Levorato
    I. Burago and M. Levorato, Cloud-Assisted On-Sensor Observation Classification in Latency-Impeded IoT Systems, in Proceedings of the IEEE International Symposium on Information Theory (ISIT), Paris, France, Jul. 2019. Slide DeckRelease

    The combination of computation and communication constraints within the Internet of Things systems require intelligent allocation of decision making and learning processes across a network of sensing and computing devices. In this paper, we present the problem of observation selection for reactive on-sensor decision-making, where the most accurate decision rule cannot be used unaided neither at the sensor (due to limited computing power), nor in the cloud (due to high communication latency). To make time-sensitive adaptation possible in these conditions, we consider learning a decision rule that is computationally viable for on-sensor use and is continuously adjusted by the cloud using the optimal decision rule for supervision. We pose a constrained stochastic optimization problem for online learning of such instrumental on-sensor classifier, propose an algorithm for updating its parameters, and establish the conditions under which convergence to a local extremum is guaranteed, at least for samples of independent observations.

    @inproceedings{burago-levorato-2019-isit,
      author = {Igor Burago and Marco Levorato},
      title = {Cloud-Assisted On-Sensor Observation Classification in Latency-Impeded {IoT} Systems},
      booktitle = {Proceedings of the IEEE International Symposium on Information Theory (ISIT)},
      address = {Paris, France},
      month = jul,
      year = {2019},
      doi = {10.1109/isit.2019.8849760},
    }
  • Igor Burago
    Marco Levorato
    I. Burago and M. Levorato, Randomized Edge-Assisted On-Sensor Information Selection for Bandwidth-Constrained Systems, in Proceedings of the Asilomar Conference on Signals, Systems, and Computers (ACSSC), Pacific Groove, California, Oct. 2018. PosterSlide DeckRelease

    The problem of intelligent information selection in the Internet-of-Things systems with limited computational and communication resources is studied. One distinctive property of such systems is the clash of the computational complexity of the desired selection procedure and the low throughput of the wireless links between the devices acquiring information (sensors) and processing it (edge and cloud computing servers). To adaptively resolve that conflict, we propose a stochastic optimization algorithm for edge-assisted online learning of the optimal on-sensor observation classification and transmission decision rules. Using the stochastic Lyapunov function method, we prove that the resulting adaptive procedure can be used to adjust the parameters of the two local decision rules to asymptotically satisfy the constraint on channel access probability and to minimize the expected classification error.

    @inproceedings{burago-levorato-2018-acssc,
      author = {Igor Burago and Marco Levorato},
      title = {Randomized Edge-Assisted On-Sensor Information Selection for Bandwidth-Constrained Systems},
      booktitle = {Proceedings of the Asilomar Conference on Signals, Systems, and Computers (ACSSC)},
      address = {Pacific Groove, California},
      month = oct,
      year = {2018},
      doi = {10.1109/acssc.2018.8645182},
    }
  • Igor Burago
    Davide Callegaro
    Marco Levorato
    Sameer Singh
    I. Burago, D. Callegaro, M. Levorato, and S. Singh, Intelligent Data Filtering in Constrained IoT Systems, in Proceedings of the Asilomar Conference on Signals, Systems, and Computers (ACSSC), Pacific Groove, California, Oct. – Nov. 2017. Release

    The expansion of complex autonomous sensing and control mechanisms in the Internet-of-Things systems clashes with constraints on computation and wireless communication resources. In this paper, we propose a framework to address this conflict for applications in which resolution using a centralized architecture with a general-purpose compression of observations is not appropriate. Three approaches for distributing observation detection workload between sensing and processing devices are considered for sensor systems within wireless islands. Each of the approaches is formulated for the shared configuration of a sensor-edge system, in which the network structure, observation monitoring problem, and machine learning-based detector implementing it are not modified. For every approach, a high-level strategy for realization of the detector for different assumptions on the relation between its complexity and the system’s constraints is considered. In each case, the potential for the constraints’ satisfaction is shown to exist and be exploitable via division, approximation, and delegation of the detector’s workload to the sensing devices off the edge processor. We present examples of applications that benefit from the proposed approaches.

    @inproceedings{burago-callegaro-2017-acssc,
      author = {Igor Burago and Davide Callegaro and Marco Levorato and Sameer Singh},
      title = {Intelligent Data Filtering in Constrained {IoT} Systems},
      booktitle = {Proceedings of the Asilomar Conference on Signals, Systems, and Computers (ACSSC)},
      address = {Pacific Groove, California},
      month = oct # {--} # nov,
      year = {2017},
      doi = {10.1109/acssc.2017.8335485},
    }
  • Korosh Vatanparvar
    Sina Faezi
    Igor Burago
    Marco Levorato
    Mohammad Abdullah Al Faruque
    K. Vatanparvar, S. Faezi, I. Burago, M. Levorato, and M. A. Al Faruque, Driving Behavior Modeling and Estimation for Battery Optimization in Electric Vehicles: Work-in-Progress, in Proceedings of the IEEE/ACM/IFIP International Conference on Hardware/Software Codesign and System Synthesis Companion (CODES), Seoul, Republic of Korea, Oct. 2017. Release

    Battery and energy management methodologies such as automotive climate controls have been proposed to address the design challenges of driving range and battery lifetime in Electric Vehicles (EV). However, driving behavior estimation is a major factor neglected in these methodologies. In this paper, we propose a novel context-aware methodology for estimating the driving behavior in terms of future vehicle speeds that will be integrated into the EV battery optimization. We implement a driving behavior model using a variation of Artificial Neural Networks (ANN) called Nonlinear AutoRegressive model with eXogenous inputs (NARX). We train our novel context-aware NARX model based on historical behavior of real drivers, their recent driving reactions, and the route average speed retrieved from Google Maps in order to enable driver-specific and self-adaptive driving behavior modeling and long-term estimation. Our methodology shows only 12% error for up to 30-second speed prediction which is improved by 27% compared to the state-of-the-art. Hence, it can achieve up to 82% of the maximum energy saving and battery lifetime improvement possible by the ideal methodology where the future vehicle speed is known.

    @inproceedings{vatanparvar-faezi-2017-codes,
      author = {Korosh Vatanparvar and Sina Faezi and Igor Burago and Marco Levorato and Mohammad Abdullah {Al~Faruque}},
      title = {Driving Behavior Modeling and Estimation for Battery Optimization in Electric Vehicles: Work-in-Progress},
      booktitle = {Proceedings of the IEEE/ACM/IFIP International Conference on Hardware/Software Codesign and System Synthesis Companion (CODES)},
      address = {Seoul, Republic of Korea},
      month = oct,
      year = {2017},
      doi = {10.1145/3125502.3125542},
    }
  • Igor Burago
    Marco Levorato
    Aakanksha Chowdhery
    I. Burago, M. Levorato, and A. Chowdhery, Bandwidth-Aware Data Filtering in Edge-Assisted Wireless Sensor Systems, in Proceedings of the IEEE International Conference on Sensing, Communication and Networking (SECON), San Diego, California, Jun. 2017. Release

    By placing processing-capable devices at the edge of local wireless access networks, Edge Computing architectures have been recently proposed to connect mobile devices to computational power through a one-hop low-latency wireless link. In this paper, we propose a new design where edge assistance is used to control local data filtering at the mobile devices in bandwidth and energy constrained systems. We focus on real-time monitoring applications, where the video input from mobile devices is processed to centrally detect and recognize objects. The edge processor controls the activation and deactivation of local classifiers implemented by the mobile devices to remove useless portions of video frames. The objective is to adapt the video stream to time-varying bandwidth constraints, while minimizing the additional energy consumption introduced by local processing. To this end, an optimization problem is formulated for a loss function embodying the balance between the risk of violating the available bandwidth and the cost of overly-conservative data filtering. The edge assists the local decision by extracting parameters of the video, such as density of objects of interest in a frame, which influence the output of the sensor. Numerical results, obtained by performing a measurement campaign based on a real implementation, illustrate the tension between energy and bandwidth use for a Haar feature-based cascade classifier.

    @inproceedings{burago-levorato-2017-secon,
      author = {Igor Burago and Marco Levorato and Aakanksha Chowdhery},
      title = {Bandwidth-Aware Data Filtering in Edge-Assisted Wireless Sensor Systems},
      booktitle = {Proceedings of the IEEE International Conference on Sensing, Communication and Networking (SECON)},
      address = {San Diego, California},
      month = jun,
      year = {2017},
      doi = {10.1109/sahcn.2017.7964938},
    }
  • Igor Burago
    Marco Levorato
    Sameer Singh
    I. Burago, M. Levorato, and S. Singh, Semantic Compression for Edge-Assisted Systems, in Proceedings of the Information Theory and Applications Workshop (ITA), San Diego, California, Feb. 2017. Release

    A novel semantic approach to data selection and compression is presented for the dynamic adaptation of IoT data processing and transmission within “wireless islands”, where a set of sensing devices (sensors) are interconnected through one-hop wireless links to a computational resource via a local access point. The core of the proposed technique is a cooperative framework where local classifiers at the mobile nodes are dynamically crafted and updated based on the current state of the observed system, the global processing objective and the characteristics of the sensors and data streams. The edge processor plays a key role by establishing a link between content and operations within the distributed system. The local classifiers are designed to filter the data streams and provide only the needed information to the global classifier at the edge processor, thus minimizing bandwidth usage. However, the better the accuracy of these local classifiers, the larger the energy necessary to run them at the individual sensors. A formulation of the optimization problem for the dynamic construction of the classifiers under bandwidth and energy constraints is proposed and demonstrated on a synthetic example.

    @inproceedings{burago-levorato-2017-ita,
      author = {Igor Burago and Marco Levorato and Sameer Singh},
      title = {Semantic Compression for Edge-Assisted Systems},
      booktitle = {Proceedings of the Information Theory and Applications Workshop (ITA)},
      address = {San Diego, California},
      month = feb,
      year = {2017},
      doi = {10.1109/ita.2017.8023457},
    }
  • Igor Burago
    Daniel Lowd
    I. Burago and D. Lowd, Automated Attacks on Compression-Based Classifiers, in Proceedings of the ACM Workshop on Artificial Intelligence and Security (AISec), Denver, Colorado, Oct. 2015, pp. 69 – 80. Release

    Methods of compression-based text classification have proven their usefulness for various applications. However, in some classification problems, such as spam filtering, a classifier confronts one or many adversaries willing to induce errors in the classifier’s judgment on certain kinds of input. In this paper, we consider the problem of finding thrifty strategies for character-based text modification that allow an adversary to revert classifier’s verdict on a given family of input texts. We propose three statistical statements of the problem that can be used by an attacker to obtain transformation models which are optimal in some sense. Evaluating these three techniques on a realistic spam corpus, we find that an adversary can transform a spam message (detectable as such by an entropy-based text classifier) into a legitimate one by generating and appending, in some cases, as few additional characters as 11% of the original length of the message.

    @inproceedings{burago-lowd-2015-aisec,
      author = {Igor Burago and Daniel Lowd},
      title = {Automated Attacks on Compression-Based Classifiers},
      booktitle = {Proceedings of the ACM Workshop on Artificial Intelligence and Security (AISec)},
      pages = {69--80},
      address = {Denver, Colorado},
      month = oct,
      year = {2015},
      doi = {10.1145/2808769.2808778},
    }
  • Igor Burago
    Igor Shevchenko
    I. Burago and I. Shevchenko, Automatic Generation of Enumeration Problems, in Proceedings of the First Russia and Pacific Conference on Computer Technology and Applications, Vladivostok, Russia, Sep. 2010.
    @inproceedings{burago-shevchenko-2010-rpc,
      author = {Igor Burago and Igor Shevchenko},
      title = {Automatic Generation of Enumeration Problems},
      booktitle = {Proceedings of the First Russia and Pacific Conference on Computer Technology and Applications},
      address = {Vladivostok, Russia},
      month = sep,
      year = {2010},
    }

Selected Presentations

  • Igor Burago
    Marco Levorato
    I. Burago and M. Levorato, Edge-Assisted On-Sensor Information Selection for Bandwidth-Constrained Systems, presented at the Information Theory and Applications Workshop (ITA), San Diego, California, Feb. 2018.
    @inproceedings{burago-levorato-2018-ita,
      author = {Igor Burago and Marco Levorato},
      title = {Edge-Assisted On-Sensor Information Selection for Bandwidth-Constrained Systems},
      booktitle = {Information Theory and Applications Workshop (ITA)},
      address = {San Diego, California},
      month = feb,
      year = {2018},
      intype = {presented at the},
    }
  • Igor Burago
    Marco Levorato
    I. Burago and M. Levorato, Parametric Identification via Likelihood Maximization in Wireless Networks, presented at the Information Theory and Applications Workshop (ITA), San Diego, California, Feb. 2016.
    @inproceedings{burago-levorato-2016-ita,
      author = {Igor Burago and Marco Levorato},
      title = {Parametric Identification via Likelihood Maximization in Wireless Networks},
      booktitle = {Information Theory and Applications Workshop (ITA)},
      address = {San Diego, California},
      month = feb,
      year = {2016},
      intype = {presented at the},
    }
  • Igor Burago
    Georgy Moiseenko
    Olga Vasik
    Igor Shevchenko
    I. Burago, G. Moiseenko, O. Vasik, and I. Shevchenko, Federating Metadata Collections on Monitoring of the North Pacific, in Proceedings of the Second International Symposium “Effects of Climate Change on the Worlds Oceans”, Yeosu, Korea, May 2012.
    @inproceedings{burago-moiseenko-2012-eccwo,
      author = {Igor Burago and Georgy Moiseenko and Olga Vasik and Igor Shevchenko},
      title = {Federating Metadata Collections on Monitoring of the {North} {Pacific}},
      booktitle = {Proceedings of the Second International Symposium ``Effects of Climate Change on the Worlds Oceans''},
      address = {Yeosu, Korea},
      month = may,
      year = {2012},
    }
  • Igor Burago
    Igor Shevchenko
    I. Burago and I. Shevchenko, Automatic Generation of Problems Using the Method of Constraint Propagation, in Proceedings of the Academician E. V. Zolotov Far Eastern Mathematical School-Seminar “Fundamental Problems of Mathematics and Information Sciences”, Khabarovsk, Russia, Jun. 2009, in Russian.
    @inproceedings{burago-shevchenko-2009-zolotov,
      author = {Igor Burago and Igor Shevchenko},
      title = {Automatic Generation of Problems Using the Method of Constraint Propagation},
      booktitle = {Proceedings of the Academician E. V. Zolotov Far Eastern Mathematical School-Seminar ``Fundamental Problems of Mathematics and Information Sciences''},
      note = {in Russian},
      address = {Khabarovsk, Russia},
      month = jun,
      year = {2009},
    }

Theses

  • Igor Burago
    I. Burago, Automated Attacks on Compression-Based Classifiers, Master’s thesis, University of Oregon, Eugene, Oregon, Jun. 2014.
    @mastersthesis{burago-2014-uo,
      author = {Igor Burago},
      title = {Automated Attacks on Compression-Based Classifiers},
      school = {University of Oregon},
      address = {Eugene, Oregon},
      month = jun,
      year = {2014},
      url = {https://scholarsbank.uoregon.edu/xmlui/handle/1794/18439},
    }
  • Igor Burago
    I. Burago, Automatic Domain-Structure Modeling Based on Testing Statistics, Master’s thesis, Far Eastern Federal University, Vladivostok, Russia, Jun. 2011, in Russian.
    @mastersthesis{burago-2011-fefu,
      author = {Igor Burago},
      title = {Automatic Domain-Structure Modeling Based on Testing Statistics},
      note = {in Russian},
      school = {Far Eastern Federal University},
      address = {Vladivostok, Russia},
      month = jun,
      year = {2011},
    }