PhD thesis 

Nonparametric recovery of signals with unknown local structure

Our goal is to explore the properties of the adaptive nonlinear filtering in the problem of signal and image recovery from indirect (incomplete and blurry) observations. First, we study the statistical properties of the proposed algorithms under various observation and signal scenarios. Second, we aim to devise fast adaptive filter implementation through iterative saddle-point optimization allowing to treat efficiently large-scale data.


Conference Papers


Structure-blind signal recovery  
D. Ostrovsky, Z. Harchaoui, A. Juditsky, A. Nemirovski 
Neural Information Processing Systems (NIPS) 2016
    
[pdf]
    
[arXiv] 

Adaptive recovery of signals by convex optimization
 
 
Z. Harchaoui, A. Juditsky, A. Nemirovski, D. Ostrovsky
Conference on Learning Theory (COLT) 2015
    
[pdf]


Journal Publications

 
Y. Golubev, D. Ostrovsky
Concentration inequalities for the exponential weighting method
 
Mathematical Methods of Statistics, vol. 23, no. 1, pp. 20-37, 2014
    
[Springer]

 
Talks and Presentations

Structure-blind signal recovery
NIPS 2016, Barcelona    [poster]

Adaptive filtering by convex optimization
PGMO Days 2016, Paris    [slides]

Adaptive recovery of  signals by convex optimization
COLT 2015, Paris    [slides]    [poster]

Near-optimal pointwise denoising by convex optimization
Statlearn 2015, Grenoble    [slides]

Concentration inequalities for the exponential weighting method
56-th Annual MIPT Conference (in Russian)    [slides]

 

MSc, 2014 (in Russian) 
Concentration inequalities for the exponential weighting method
 
Moscow Institute of Physics and Technology, Department of Electrical Engineering and Cybernetics
  
 
[pdf]
Adviser:  Pr. Yuri Golubev
  
BSc, 2012 (in Russian)
Analytical study of NHDP link management protocol 
Moscow Institute of Physics and Technology, Department of Electrical Engineering and Cybernetics    [pdf] 
Adviser:  Pr. Andrey Lyakhov 
  
 
E. 
Khorov, A. Krasilov, A. Lyakhov, and D. Ostrovsky 
Dynamic resource allocation for MCCA-based streaming in Wi-Fi mesh networks 
Wireless Access Flexibility 2013
    [Springer]
  
E. Khorov, A. Kiryanov, A. Lyakhov, D. Ostrovsky 
Analytical study of 
neighborhood discovery and link management in OLSR

IFIP Wireless Days 2012    [pdf]
  
A. Krasilov, A. Lyakhov, D. Ostrovsky, E. Khorov 
A dynamic channel reservation method for multimedia streaming in Wi-Fi Mesh networks
Automation and Remote Control, vol. 74, no. 9, pp. 1460-1473, 2013    [Springer] 
  
A. Lyakhov, D. Ostrovsky, E. Khorov
Analytical study of the quality of links established by the neighborhood discovery protocol 
Journal of Communications Technology and Electronics, vol. 57, no. 12, pp. 1314-1321, 2012    [Springer] 
  

* As an undergraduate, I was a research assistant at Network Protocols Research Lab of IITP RAS (Kharkevich Institute)