Main Page
CV
 Publications
 Research
Nuri Denizcan Vanli
M.S. Student, Department of Electrical and Electronics Engineering
Bilkent University, Ankara 06800, Turkey
E-mail: vanli@ee.bilkent.edu.tr		  

He conducts research broadly, but not limited to, in the following topics:

Machine Learning

In many real life applications, the volume of data to be handled has significantly increased due to the recent developments in information technologies. Aside from the difficulties of storing such massive amounts of data, the processing of data usually has to be done in real-time. Hence, the need for machine learning algorithms that can process the data on-the-fly with significantly low computational complexities is steadily growing. In this context, there exist significant practical and theoretical difficulties to sequential learning, since there is usually no or little knowledge about the statistical properties of the underlying signals or systems involved. Furthermore, the classical robust methods that are resilient to such challenges are overly conservative and usually static such that they provide profoundly inferior results on the average, deeming them practically useless.

In order to provide robust adaptive methods that also perform satisfactorily in real life applications, we introduce sequential learning algorithms that are mathematically guaranteed to work uniformly for all possible signals without any explicit or implicit statistical assumptions on the underlying signals or systems. We construct adaptive algorithms that sequentially perform as well as the best batch algorithm, for any signal, that had the ability to observe the entire data in advance. Unlike the state-of-the-art methods in the literature, our algorithms can tackle problems with non-stationarity in data modeling by sequentially learning the model structure. In this manner, without introducing any ad-hoc assumptions or parameters, we introduce algorithms that directly minimize any desired loss measure in a strong deterministic sense. Hence, our results are guaranteed to hold, not on the average, but in an individual sequence manner.

Publications on this topic include:

  1. N. D. Vanli and S. S. Kozat, "A Comprehensive Approach to Universal Piecewise Nonlinear Regression Based on Trees," IEEE Transactions on Signal Processing, vol. 62, no. 20, pp. 5471-5486, Oct. 2014. (ieeexplore) (local copy)
        
  2. N. D. Vanli and S. S. Kozat, "A Unified Approach to Universal Prediction: Generalized Upper and Lower Bounds," IEEE Transactions on Neural Networks and Learning Systems, accepted 2014. (ieeexplore) (local copy)
        
  3. N. D. Vanli, M. O. Sayin, and S. S. Kozat, "Predicting Nearly As Well As the Optimal Twice Differentiable Regressor," Machine Learning, 2014. (arxiv)
        
  4. N. D. Vanli, M. O. Sayin, H. Yilmaz, T. Goze, and S. S. Kozat, "Energy Consumption Forecasting via Order Preserving Pattern Matching," to appear in IEEE Global Conference on Signal and Information Processing (GlobalSIP) , 2014. (local copy)
        
  5. N. D. Vanli, M. O. Sayin, S. Ergut, and S. S. Kozat, "Piecewise Nonlinear Regression via Decision Adaptive Trees," European Signal Processing Conference (EUSIPCO), Sep. 1-5, 2014. (eurasip) (local copy)
        
  6. N. D. Vanli, M. O. Sayin, S. Ergut, and S. S. Kozat, "Comprehensive Lower Bounds on Sequential Prediction," European Signal Processing Conference (EUSIPCO), Sep. 1-5, 2014. (eurasip) (local copy)
        
  7. N. D. Vanli and S. S. Kozat, "Sequential Nonlinear Regression via Context Trees," IEEE Signal Processing and Communications Applications Conference (SIU), pp. 1865-1868, April 23-25, 2014. (ieeexplore) (local copy)
        

Distributed Processing

Recently, we have experienced a dramatic growth in the capabilities of sensors and distributed sensor networks. Due to advances in device technologies, we now have sensors that can, not only collect data, but also process and communicate it among neighbors. The rapid increase in the processing power and decline in device dimensions of these sensors opened a wide range of new application areas for distributed processing, including mobile device networks, robot swarms and wearable electronics. We can tremendously increase performance using the collective intelligence and processing power of these advance units that can gather, process and communicate data, replacing the traditional sensors that only communicate observations to a central processing unit.

There exists important theoretical and practical challenges to effectively use adaptive algorithms in the state-of-the-art distributed networks under real life conditions. The communication and energy consumption constraints, rapidly changing topologies and application requirements impede realistic usage of adaptive signal processing techniques. We aim to (1) develop, as the first time, novel distributed adaptive signal processing algorithms specifically designed to effectively operate under real life communication constraints, (2) demonstrate the effectiveness and efficiency of these approaches in different applications and prove these results mathematically, (3) provide algorithms with security guarantees in order to work under strict military and sensitive commercial specifications, and finally (4) construct a practical distributed adaptive signal processing framework by defining new cost measures and design methodologies that include realistic communication and power constraints.

Publications on this topic include:

  1. N. D. Vanli, M. O. Sayin, and S. S. Kozat, "Stochastic Subgradient Algorithms for Strongly Convex Optimization over Distributed Networks," IEEE Transactions on Signal and Information Processing over Networks, 2015. (arxiv)
        
  2. N. D. Vanli, M. O. Sayin, and S. S. Kozat, "Comments on "Distributed Autonomous Online Learning: Regrets and Intrinsic Privacy-Preserving Properties"," IEEE Transactions on Knowledge and Data Engineering, 2014. (local copy)
        
  3. M. O. Sayin, N. D. Vanli, T. Goze, and S. S. Kozat, "Communication Efficient Channel Estimation Over Distributed Networks," to appear in IEEE Global Conference on Signal and Information Processing (GlobalSIP) , 2014. (local copy)
        
  4. M. O. Sayin, N. D. Vanli, and S. S. Kozat, "Logarithmic Regret Bound Over Diffusion Based Distributed Estimation," IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), pp. 8287-8291, May 4-9, 2014. (ieeexplore) (local copy)
        

Adaptive Filtering and Control Theory

The field of adaptive signal processing experienced dramatic growth due to the proliferation of new and exciting applications ranging from Internet, wireless communications to multimedia and quantitative finance. Adaptive systems have become an integral part of information and telecommunications industries as a result of advances in device technology. As the range of environments that these signal processing applications are expected to work are increasing, there is now a greater need for adaptive algorithms that can operate efficiently in the presence of a wide range of environmental uncertainties and volatility with relatively low computationally complexity. In this context, there exists significant practical and theoretical difficulties to adaptive signal processing.

Adaptive filtering applications such as channel equalization, noise removal or echo cancellation utilize a certain statistical measure of the error signal given by the difference between the desired signal and the estimation output. Usually, the mean square error is used as the cost function due to its mathematical tractability and relative ease of analysis. The performance of the least-squares algorithms degrades severely when the input and desired signal pairs are perturbed by impulsive interferences, e.g., in applications involving high power noise signals. To this end, different powers of the error are commonly used as the cost function in order to provide stronger convergence or steady-state performance than the least-squares algorithms under certain settings. In the computational learning theory related to the "competitive" or the "regret" based approaches, the well known adaptive or online learning algorithms are "stabilized" or "improved" by using a relative cost measure, i.e., the regret. Inspiring from these recent developments, we mitigate the stability or convergence issues of the well known adaptive algorithms by introducing a relative cost measure.

Publications on this topic include:

  1. M. O. Sayin, N. D. Vanli, and S. S. Kozat, "A Novel Family of Adaptive Filtering Algorithms Based on The Logarithmic Cost," IEEE Transactions on Signal Processing, vol. 62, no. 17, pp. 4411-4424, Sep. 2014. (ieeexplore) (local copy)
        
  2. M. O. Sayin, N. D. Vanli, and S. S. Kozat, "Improved Convergence Performance of Adaptive Algorithms with Logarithmic Cost," IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), pp. 4513-4517, May 4-9, 2014. (ieeexplore) (local copy)
        
  3. M. O. Sayin, N. D. Vanli, and S. S. Kozat, "Robust Set-Membership Filtering Algorithms Against Impulsive Noise," IEEE Signal Processing and Communications Applications Conference (SIU), pp. 1551-1554, April 23-25, 2014. (ieeexplore) (local copy)
        

Signal Processing for Communications

In many practical communications systems, the communications channel is usually unknown and possibly time-varying. The efficient communication over such systems requires the estimation of the channel, where this estimate is usually subject to distortion. In this context, in order to provide reliable communications, robust estimation algorithms are needed. However, the classical robust methods that are resilient to such distortions are overly conservative, hence provide unacceptable results on the average.

In order to present robust algorithms without sacrificing from the average estimation performance, we propose novel robust approaches that minimize a worst case regret that is defined as the difference between the estimation error and the smallest attainable estimation error with an LS or MMSE estimator. In this sense, we seek an estimator, whose performance is as close as possible to that of the optimal estimator for any possible perturbation. By this formulation, we alleviate the highly conservative nature of the classical robust methods and introduce robust algorithms that provide satisfactory performance on the average, unlike the classical approaches.

Publications on this topic include:

  1. N. D. Vanli, M. A. Donmez, and S. S. Kozat, "Robust Least Squares Methods Under Bounded Data Uncertainties," Digital Signal Processing, vol. 36, pp. 82-92, Jan. 2015. (elsevier) (local copy)
        
  2. B. Dulek, N. D. Vanli, S. Gezici, and P. K. Varshney, "Optimum Power Randomization for the Minimization of Outage Probability," IEEE Transactions on Wireless Communications, vol. 12, no. 9, pp. 4627-4637, Sep. 2013. (ieeexplore) (local copy)
        
  3. S. Bayram, N. D. Vanli, B. Dulek, I. Sezer, and S. Gezici, "Optimum Power Allocation for Average Power Constrained Jammers in the Presence of Non-Gaussian Noise," IEEE Communications Letters, vol. 16, no. 8, pp. 1153-1156, Aug. 2012. (ieeexplore) (local copy)
        
  4. N. D. Vanli, M. A. Donmez, and S. S. Kozat, "Robust Regularized Least Squares Estimation in the Presence of Bounded Data Uncertainties," IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), pp. 5637-5640, May 4-9, 2014. (ieeexplore) (local copy)
        
  5. N. D. Vanli, M. O. Sayin, and S. S. Kozat, "Competitive Linear MMSE Estimation Under Structured Data Uncertainties," IEEE Signal Processing and Communications Applications Conference (SIU), pp. 1861-1864, April 23-25, 2014. (ieeexplore) (local copy)
        
  6. B. Dulek, N. D. Vanli, and S. Gezici, "Convexity Properties of Outage Probability under Rayleigh Fading," IEEE Signal Processing and Communications Applications Conference (SIU), 2012. (ieeexplore) (local copy)