I am a PhD Candidate in the Department of Computer Science
of the University of California, Santa Barbara, working as a Graduate
Student Researcher at the Databases,
Data Mining, and Bioinformatics Lab (DBL), and being advised by
Ambuj K. Singh. My research
interests are machine learning and data mining, combinatorial algorithms,
linear algebra, and their application to the analysis of dynamic
graphs, such as online social networks. Prior to joining DBL, I had
worked in high-performance computing
and spent several years as a
software engineer in industry. I have obtained my Master's degree in
Applied Mathematics and Computer Science with an emphasis on Numerical
Tula State University,
Russia in 2008, having been advised
by Valeriy Ivanov.
Amelkin V., Bullo F., Singh A.K.
"Polar Opinion Dynamics in Social Networks",
Submitted to IEEE Transactions on Automatic Control (May, 2016)
Amelkin V., Askarisichani O., Kim Y.J., Singh A.K., Malone T.W.
"Dynamics of Collective Performance
in Collaboration Networks",
INSNA Sunbelt (2016),
Amelkin V., Ng E.G.
"Parallel Communication Analysis for Sparse Cholesky
LBNL (2012), [report]
Amelkin V., Ivanov A., "Fejér problem for polynomials on
a two-dimensional sphere",
Modern Problems of Mathematics,
Mechanics, Computer Science; Tula State University (2008)
— a fast transportation problem solver for MATLAB, based on
Andrew Goldberg's CS2
implementation of Goldberg-Tarjan's min-cost network flow algorithm.
It works much faster than MATLAB's linprog or CPLEX' general-purpose
— multiple MATLAB implementations of Dijkstra's single-source shortest
path algorithm for sparse networks: the binary heap-based implementation, due
to David Bindel;
the radix heap-based implementation; and the implementation
based on an improved version of Dial's algorithm.
— a GNU/Linux and Mac OS X port of Complex Network Package 1.6 — a graph library for MATLAB.
— a MATLAB API to AlchemyAPI's text sentiment quantification
web-service. Academic users can get 30k transactions per day for free.
The sentiment quantification accuracy for Sentiment140's training set of tweets was around 70%.
— materials from my Major Area Exam.