Hyeryung Jang

Contact me: hrjang357 at gmail

Bio, Long CV

I am currently a research associate in the Department of Informatics of King's College London (KCL), London, England, United Kingdom, working with Prof. Osvaldo Simeone. I worked as a post-doctoral researcher with Prof. Yung Yi at Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of Korea. I received Ph.D. in Electrical Engineering from KAIST in 2017, working with Prof. Yung Yi and Prof. Jinwoo Shin. I received M.S. and B.S. in Electrical Engineering from KAIST in 2012 and 2010, respectively.

My research interests include network economics, game theory, analysis of communication systems, and machine learning, especially applying learning of graphical models to communication systems. In particular, my research interests lie in mathematical modeling and analysis with a specific focus on learning graphical models and stochastic approximation among others. Recently, my research interests focus on the area of Networked Machine Learning which includes machine learning (supervised, unsupervised and reinforcement learning) in large-scale communication networks. Here are examples of topics covered by my recent research works:

• Machine Learning in Graphical Models - Ising model, Parameter learning [ISIT2017], Structure learning [MOBIHOC2018]

• Distributed Optimization in Communication Networks - optimal CSMA [INFOCOM2014, TWC2017], Coordination maximization [MOBIHOC2016, TCNS2018]

• Game Theory and Economics in Networks - Network economics, Pricing of ISPs [JSAC2017, SDP2013]

Publication

You can also find my papers listed in Google Scholar, or dblp.

Journal

  • [J3] Hyeryung Jang, Jinwoo Shin, and Yung Yi, "Simulation-based Distributed Coordination maximization over Networks,” to appear at IEEE Transactions on Control of Network Systems, 2018. (Part of this work has been published MOBIHOC 2016) (ArXiv preprint version is available at arXiv:1809.04972)
  • [J2] Hyeryung Jang, Se-Young Yun, Jinwoo Shin, and Yung Yi, "Game Theoretic Perspective of Optimal CSMA,” IEEE Transactions on Wireless Communications, vol.17, no.1, pp.194-209, Jan. 2018. (A shorter version has been published INFOCOM 2014)
  • [J1] Hyojung Lee, Hyeryung Jang, Jeong-woo Cho, and Yung Yi, "Traffic Scheduling and Revenue Distribution among Providers in the Internet: Trade-offs and Impacts,” IEEE Journal on Selected Areas in Communications (JSAC) Special Issue on Game Theory for Networks, vol.35, no.2, pp.421-431, Feb. 2017. (Part of this work has been published INFOCOM SDP workshop 2013)

Conference

Thesis

  • [T2] Hyeryung Jang, "Optimization and Learning of Graphical Models: A Stochastic Approximation Approach," Ph.D. Thesis in Electrical Engineering from Korea Advanced Institute of Science and Technology (KAIST), 2017. (Advisor: Prof. Yung Yi, Co-Advisor: Prof. Jinwoo Shin)
    • Abstract: This thesis mainly addresses the problem of optimization and learning in graphical models via stochastic approximation theory. First, in various multi-agent networked environments, the system can benefit from coordinating actions of two interacting agents at some cost of coordination, where a primary goal is to develop distributed algorithms maximizing the coordination effect. Examples include a wireless sensor networks with duty cycling, where a sensor node consumes a certain amount of energy when it is awake, but a coordinated operation of sensors enables some meaningful tasks, e.g., sensed data forwarding. Such pair-wise coordinations and node-wise costs in the network can be captured by graphical model framework, which becomes the problem of finding the optimal graph parameter. In this thesis, I present various distributed algorithms that require only one-hop message passing and locally-observed information, which can be interpreted based on either Lagrange duality theory or game theory framework. The proposed algorithms are motivated by a stochastic approximation method that runs a Markov chain simulation incompletely over time, but provably guarantees the convergence to the optimal solution. Second, I discuss the problem of parameter learning in graphical models having latent variables, where the standard approach, i.e., Expectation Maximization algorithm, is computationally intractable for high dimensional data, in both expectation and maximization steps. Since the substitution of one step to a faster surrogate for combating against intractability can often cause failure in convergence, I propose a new learning algorithm (called Adiabatic Persistent Contrastive Divergence), which runs a few cycles of Markov chains in both steps. Using multi-time-scale stochastic approximation theory, the proposed algorithm provably ensures convergence to a correct optimum, and moreover, I demonstrate the theoretical findings through extensive experiments with synthetic data and/or real-world data sets.
  • [T1] Hyeryung Jang, "Economic Benefits of ISP-CDN and ISP-ISP Cooperation," M.S. Thesis in Electrical Engineering from Korea Advanced Institute of Science and Technology (KAIST), 2012. (Advisor: Prof. Yung Yi)

MISC

History

Work Experiences

  • 2018.03~Current: Research Associate, Centre for Telecommunications Research, Department of Informatics, King's College London, London, England, United Kingdom (Host: Prof. Osvaldo Simeone)
  • 2017.03~2018.02: Post-doctoral Researcher, BrainKorea21 Plus, Information & Electronics Research Institute, KAIST, Republic of Korea (Host: Prof. Yung Yi)
  • 2015.06~2015.10: Research Intern, Center of Non-Linear Studies, Los Alamos National Laboratory, NM, United State (Host: Dr. Michael Chertkov)

Project Experiences

  • 2018.03~Current: FOG-aided wireless networks for communication, cacHing and cOmputing: theoRetical and algorithmic fouNdations (FOGHORN), European Research Council (ERC)
    • Develop fundamental theoretical insights, via network information theory, communication theory and machine learning, on the optimal performance and operation of fog-aided wireless networks
  • 2017.07~2018.03: Research on Learning-based Service Improvement Framework in Large-scale Online Request System, Naver, Republic of Korea
    • Design and analyze Reinforcement Learning based service improvement framework in large-scale online request system
    • Implement deep reinforcement learning algorithm for service improvement in large-scale online request system
  • 2016.04~2018.02: Versatile network System Architecture for Multi-dimensional Diversity, Ministry of Science, ICT & Future Planning (MSIP), Republic of Korea
    • Research on edge, cloud, fog networks
  • 2014.10~2015.06: Bosch-KAIST Smart Car Project: Look Ahead: Shared Sensing for Cooperative Cars, Bosch, German - Korea
    • Design and implement MAC protocol for real-time video transmission between vehicles
  • 2013.11~2016.10: Research on Horizontal and Vertical Decoupling in Big Wireless Networks: Theory and Implementation, National Research Foundation of Korea (NRF) grant funded by the Ministry of Science, ICT & Future Planning (MSIP), Republic of Korea
    • Research on distributed MAC protocol design for CSMA-based wireless networks (from optimization and game theory perspective)
  • 2011.06~2011.10: Research on Smart Network based B2B2C Service Modeling and Economic Analysis, Korea Telecom, Republic of Korea
    • Research on analysis of economic value of pricing between Internet Service Providers, Content Providers, and users
  • 2010.03~2012.02: Modeling of Network Architecture Considering Network Economics & Efficient Routing Method on Delay Tolerant Network, National Research Foundation of Korea (NRF) grant funded by the Ministry of Knowledge Economy, Republic of Korea
    • Design efficient routing protocol for delay tolerant network

Presentation Experiences

  • "Adiabatic Persistent Contrastive Divergence Learning," at IEEE International Symposium on Information Theory (ISIT), Aachen, Germany, June 30, 2017. [slide]
  • "Optimization and Learning of Graphical Models: A Stochastic Approximation Approach," at Korea Computer Congress (KCC): Spotlight Session for Young Women Scholars, Jeju, Republic of Korea, June 20, 2017. [slide]
  • "Distributed Coordination Maximization over Networks: A Stochastic Approximation Approach,", at ACM Mobile Ad Hoc Networking and Computing (MobiHoc), Paderborn, Germany, July 7, 2016. [slide]
  • "Distributed Learning for Utility Maximization over CSMA-based Wireless Multihop Networks," at IEEE International Conference on Computer Communications (INFOCOM), Toronto, Canada, April 29, 2014. [slide]
  • "On the Interaction between ISP Revenue Sharing and Network Neutrality," at ACM International Conference on emerging Networking Experiments and Technologies (CoNEXT) Student workshop, Philadelphia, USA, November 30, 2010. [poster]

Major Courses Taken and Technical Skills

Theory

  • Electrical Engineering: Optimization in Communication Networks, Economics in Communication Networks, Analysis of Complex Networks, Statistical Learning Theory, Artificial Intelligence and Machine Learning, Introduction to Big Data
  • Mathematical Science: Mathematical analysis, Lebesgue Integrals, Queueing Theory with Applications, Probability Theory

Systems

  • Data Structure and Algorithms, Algorithms: Design and Analysis

Technical skills

  • Matlab (for parameter learning in graphical models simulations)
  • C and C++ (for optimal CSMA, distributed coordination maximization simulations)
  • Python and Tensorflow (for reinforcement learning based algorithm in large-scale online request system simulations)

Teaching Experiences

  • Teaching Assistant, Optimization in Communication Networks (EE650), KAIST, Spring 2016, Spring 2013
  • Teaching Assistant, Programming Structures for Electrical Engineering (EE209), KAIST, Fall 2014
  • Teaching Assistant, Computer Networks (EE323), KAIST, Spring 2014, Spring 2011
  • Teaching Assistant, Data Structure and Algorithms for Electrical Engineering (EE205), KAIST, Fall 2013, Fall 2011
  • Teaching Assistant, Economics in Communication Networks (EE655), KAIST, Spring 2012

Education

  • Ph.D: Electrical Engineering, KAIST, 2012.3~2017.2 (under supervision of Prof. Yung Yi and Prof. Jinwoo Shin)
  • M.S: Electrical Engineering, KAIST, 2010.3~2012.2 (under supervision of Prof. Yung Yi)
  • B.S: Electrical Engineering, KAIST, 2006.3~2010.2
  • Highschool: Korea Minjok Leadership Academy (KMLA), 2004.3~2006.2

Honors and Awards

  • BK21 Plus Scholarship, Republic of Korea, 2017
  • Qualcomm-KAIST Innovation Award, Qualcomm, 2016
  • National Scholarship, Republic of Korea, 2006~2016


Reference Available on Request

  • Professor Yung Yi: Associate Professor at the Department of Electrical Engineering, KAIST, yiyung at kaist dot edu, +82 (0) 42 350 3486
  • Professor Jinwoo Shin: Associate Professor at the Department of Electrical Engineering, KAIST, jinwoos at kaist dot ac dot kr
  • Professor Se-Young Yun: Assistant Professor at the Department of Industrial & Systems Engineering, KAIST, yunseyoung at kaist dot ac dot kr