Dmitrii Avdiukhin


Dmitrii Avdiukhin

Ph.D. student, Indiana University

Advisor: Grigory Yaroslavtsev

Resume

e-mail: davdyukh (at) iu (dot) edu



Research interests

Education

2017-current
Ph.D. student, Indiana University.
2008-2013
Specialist (5 years) degree. Saint Petersburg State University. Diploma with distinction. Thesis title: "Translation definition language for information system reengineering tools".

Publications

AISTATS 2021
(under review)
D. Avdiukhin., and G. Yaroslavtsev. "Escaping Saddle Points with Inequality Constraints via Noisy Sticky Gradient Descent"
AISTATS 2021
(under review)
D. Avdiukhin., and S. Kasiviswanathan. "Federated Learning under Arbitrary Communication Patterns"
AAAI 2021
D. Avdiukhin., S. Naumov, and G. Yaroslavtsev. "Objective-Based Hierarchical Clustering of Deep Embedding Vectors"
OPT 2020
D. Avdiukhin., and G. Yaroslavtsev. "Escaping Saddle Points with Compressed SGD"
AISTATS 2020
G. Yaroslavtsev , S. Zhou, and D. Avdiukhin. ““Bring Your Own Greedy”+Max: Near-Optimal 1/2-Approximations for Submodular Knapsack”
OPT 2019
D. Avdiukhin, G. Yaroslavtsev, and C. Jin. “Escaping Saddle Points with Inequality Constraints via Noisy Sticky Projected Gradient Descent”
VLDB 2019
D. Avdiukhin, S. Pupyrev and G. Yaroslavtsev. “Multi-Dimensional Balanced Graph Partitioning via Projected Gradient Descent”
KDD 2019
D. Avdiukhin, S. Mitrovic, G. Yaroslavtsev and S. Zhou. “Adversarially Robust Submodular Maximization under Knapsack Constraints”. Oral presentation, 9.2% acceptance rate.
INDIN 2017
D. Avdyukhin, D. Chivilikhin, G. Korneev, V. Ulyantsev and A. Shalyto. “Plant trace generation for formal plant model inference: methods and case study”
Perspectives of System Informatics 2015
E. Verbitskaia, S. Grigorev and D. Avdyukhin. “Relaxed Parsing of Regular Approximations of String-Embedded Languages”

Experience

Summer 2020
Research Intern, Amazon.
Federated Learning under weak assumptions
Summer 2019
Research Intern, Amazon, New York.
Improving accuracy and performance of graph convolutional networks
Summer 2018
Software Engineer, Pro Unlimited @ Facebook, Menlo Park.
Implementing balanced graph partitioning algorithm
2016-2017
Researcher, ITMO University.
Model generation from execution traces
2013-2016
Software Engineer. JetBrains, Saint Petersburg.
SQL dialects support
2012-2013
Software Engineer. Lanit Tercom, Saint Petersburg.
Participating in project of migration a system from SQL server to Oracle

Talks

VLDB 2019
Presenting “Multi-Dimensional Balanced Graph Partitioning via Projected Gradient Descent”
KDD 2019
Presenting “Adversarially Robust Submodular Maximization under Knapsack Constraints”
Yandex, Moscow,
Russia, 2019
"Multi-Dimensional Balanced Graph Partitioning via Projected Gradient Descent"

Fellowships

2019
Nominated for Google PhD Fellowship Program by Indiana University
2019
Nominated for Microsoft Fellowship by Indiana University

Teaching

Spring 2020
"Applied Algorithms" (graduate). Indiana University, Bloomington, CSCI-B505. Head Associate Instructor.
Spring 2019
"Data Structures" (Honors, undergraduate)}. Indiana University, Bloomington, CSCI-H343. Associate Instructor.

Competitive programming

ACM ICPC
Latest result: 19th place on world semi-final, 2014: Standings
Codeforces
Rating: 2135
Topcoder
Rating: 1784