Sanghyun Hong 홍상현 메릴랜드 대학교

Ph.D Student in Computer Science
at University of Maryland, College Park, advised by Prof. Tudor Dumitraș.

My primary interest is to solve security and privacy problems of machine learning (ML) systems. Recently, I've exposed new attack vectors in deep learning (DL) systems and proposed defense mechanisms [1, 2, 3], and characterized the internal behaviors of deep neural networks against adversarial samples [4]. Also, I've introduced security services for cloud infrastructure [5, 6], where DL systems are running.

Before moving to Maryland, I received my B.S. in Electrical Engineering and Computer Science from Seoul National University in 2015. During my military service (2010-13), I carried out projects with LG Electronics Inc. (LGE) as a lead researcher. Also, I founded Openwise Inc. (2013-15), a start-up company where I worked as a chief technology officer, and supervised research projects carried out with Samsung Advanced Institute of Technology (SAIT).

Contacts

Address: 5112 Brendan Iribe Center for Computer Science and Engineering, College Park, MD, USA [Maryland Cybersecurity Center (MC2)]
Email: shhong [at] cs [dot] umd [dot] edu

Let's Get Connected: Google Scholar | GitHub | Twitter | LinkedIn

Selected Publications


— Security and Privacy of Machine Learning
How to 0wn NAS in Your Spare Time.

Sanghyun Hong, Michael Davinroy, Yigitcan Kaya, Dana Dachman-Soled, and Tudor Dumitraș.
International Conference on Learning Representations, 2020 (ICLR).
Paper Poster Website

Terminal Brain Damage: Exposing the Graceless Degradation in Deep Neural Networks Under Hardware Fault Attacks.

Sanghyun Hong, Pietro Frigo, Yigitcan Kaya, Cristiano Giuffrida, and Tudor Dumitraș.
USENIX Security Symposium, 2019 (USENIX).
Paper Presentation Slides Website

Shallow-Deep Networks: Understanding and Mitigating Network Overthinking.

Yigitcan Kaya, Sanghyun Hong, and Tudor Dumitraș.
International Conference on Machine Learning, 2019 (ICML).
Paper Slides Poster Code Website


— Secure Cloud Infrastructures
Peek-a-Boo: Inferring Program Behaviors in a Virtualized Infrastructure without Introspections.

Sanghyun Hong, Alina Nicolae, Abhinav Srivastava, and Tudor Dumitraș.
Computer & Security, 2018 (COSE).
Paper

Go Serverless: Securing Cloud via Serverless Design Patterns.

Sanghyun Hong, Abhinav Srivastava, William Shambrook, and Tudor Dumitraș.
USENIX Workshop on Hot Topics in Cloud Computing, 2018 (HotCloud 18).
Paper Slides


— Network (Knowledge Graph) Embeddings
SENA: Preserving Social Structure for Network Embedding.

*Sanghyun Hong, *Tanmoy Chakraborty, Sungjin Ahn, Ghaith Husari, Noseong Park. (* indicates equal contribution)
ACM Conference on Hypertext and Social Media, 2017 (ACM HT).
Paper

Awards

2018: 2nd/4th Place at Research Competition for Korean Graduate Students
2017: KSEA-KUSCO Scholarships for Korean Graduate Students in the United States
2016: Summer Research Fellowship from the Department of Computer Science at UMD
2015: 2-year Dean's Fellowship from UMD for outstanding academic achievement
2015: Full 2-year Graduate Teaching Assistantship from UMD for academic excellence