Arsenii Mustafin

Arsenii Mustafin is a fourth year PhD student at Boston University, department of Computer Science. He is advised by Prof. Alex Olshevsky and Prof. Yannis Paschalidis. His main interest is Reinforcement learning, both theoretical and practical. Arsenii is also interested in other topics in machine learning, including computer vision, explainability and deep fake detection. Previously Arsenii worked on SemaFor DARPA project with BU team lead by Prof. Kate Saenko and a project on Explainable AI with Prof. Sarah Adel Bargal.

Current coursework: EC 500 A2: Robot Learning and Vision for Navigation

Completed coursework: EC700: Reinforcement learning, CS591: Deep Learning, CS542: Machine Learning, EC724: Advanced Optimization, CS565: Algorithmic Data Mining, CS537: Randomness in Computing, CS531: Advanced optimization algorithms, CS655: Computer networks.

Previously Arsenii was studying economics, and took several undergraduate and graduate-level courses on econometrics and time-series analysis. Prior to coming to BU, Arsenii has been working as a machine learning developer. Arsenii spent four years in Mainland China and speaks fluent Mandarin, in addition to fluent English and native Russian.

aam [at] bu.edu  |  CV  |  LinkedIn  |  Github

Algorithmic: Reinforcement Learning, Deep Learning, classic Machine Learning algorithms, Time Series analysis.

Software: PyTorch, Tensorflow (1&2), Linux (+git), SQL.

Arsenii completed several learning, research and practice-oriented projects on machine learning.

Closing the Gap Between SVRG and TD-SVRG with Gradient Splitting
Mustafin A., Olshevsky, A., Paschalidis, I.
2022

In the paper Arsenii and his coauthors significantly improve theoretical guarantees of SVRG method applied to TD update, show that it exhibits the same convergence as SVRG in convex optimization setting and provide theoretical guarantees for practical algorithm.

Ani-GIFs: A Benchmark Dataset for Domain Generalization of Action Recognition from GIFs
Mustafin, A.*, Jain, S.*, Lteif, D.*, Majumdar S.*, Tourni, I.*, Bargal S., Saenko K., Sclaroff S.
2022

This paper published in "Frontiers in computer science" presents a dataset for Domain Generalization problems in the video domain.

Patent: SYSTEMS AND METHODS FOR IMAGE OR VIDEO PERFORMANCE HEAT MAP GENERATION
Mustafin A., Saraee E., Hamedi J., Halloran Z.
2020

The patent application was a results of Arsenii's internship at Vizit labs, Inc, during which Arsenii played a key role in developing an explainability techinique for computer vision model.

BU CS 542 in-class kaggle competition
Arsenii Mustafin
2019

Arsenii got second place on graduate machine learning class competition where students were required to predict airbnb property listing prices without using external machine learning libraries.

MyNN project
Arsenii Mustafin
2018

Arsenii has completed this project during his work with Intel (Shanghai). In this project he has build a neural network module from scrath, using only numpy. He applied this module to implement Reinforcement learnig technics, from simple DQN and Policy Gradient, to TRPO and PPO with GAE, which back then were state of art algorithms.

Research on EAEU trade
Arsenii Mustafin
2017

In this paper Arsenii applied quantitative analysis to assess, how formation of Eurasian Custom Union and Eurasion Economic union affected trade between member countries. Difference-in-diffirence methodology and various settings of gravity model were applied, trade dynamics of member countries was compared with trade dynamics among wider set of countries - members of Commonwealth of Independent States. Research have shown, that such strong measures as custom and econimic union formation has not significantly affected trade between union members.

template