Research Interest: Graph Mining, Social Network Analysis, Stream Mining, and Autonomous Machine Learning
I am a PhD student majoring in Computer Science at Carnegie Mellon University.
I am very fortunate to be advised by Prof. Christos Faloutsos.
Before that, I was a software developer in SAP Labs Korea.
I received M.S. in Computer Science, under Prof. Jehee Lee, and B.S. in Electronic Engineering at Seoul National University.
Ph.D Computer Science
M.S. Computer Science and Engineering
B.S. Electrical and Computer Engineering
Seoul, Korea (Mar. 2006 - Feb. 2008)
- MIDAS: Microcluster-Based Detector of Anomalies in Edge Streams
Siddharth Bhatia, Bryan Hooi, Minji Yoon, Kijung Shin, and Christos Faloutsos
34th AAAI Conference on Artificial Intelligence (AAAI) 2020
- Fast and Accurate Anomaly Detection in Dynamic Graphs with a Two-Pronged Approach
Minji Yoon, Bryan Hooi, Kijung Shin, and Christos Faloutsos
SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) 2019, Alaska, USA
- Fast and Accurate Random Walk with Restart on Dynamic Graphs with Guarantees
Minji Yoon, Woojeong Jin, and U Kang
The Web Conference (WWW) 2018, Lyon, France
(Acceptance Ratio: 14.8%)
- TPA: Fast, Scalable and Accurate Method for Approximate Random Walk with Restart on Billion Scale Graphs
Minji Yoon, Jinhong Jung, and U Kang
IEEE International Conference on Data Engineering (ICDE) 2018, Paris, France
- Online Motion Puppetry for Non-human Characters
Minji Yoon and Jehee Lee
Korea Computer Graphics Society (KCGS) 2014
- Crowd simulation based on Motion Patches
Minji Yoon, Kyunglyul Hyun, and Jehee Lee
Korean Society for Precision Engineering (KSPE) 2013
Applied Scientist Intern (May. 2019 - Aug. 2019)
Research Intern (Apr. 2017 - Jun. 2018)
Software Developer (Sep. 2014 - Mar. 2017)
Research Intern (Mar. 2012 - Aug. 2012)
- Fast, Scalable and Accurate Method for Approximate Random Walk with Restart on Billion Scale Graphs, PCT/KR2017/012962 (18 Nov 2017)
Minji Yoon, Jinhong Jung, and U Kang
- Remote authentication, USA: 15/363,102 (29 Nov 2016)
Minji Yoon and Chulwon Lee
- Workload Shifting in a Database System Using Hint-Based Routing
US patent 10,162,721 (grant date Dec. 25, 2018)
Yongwook Jeong, Minji Yoon, Ian McHardy, Jeff Albion, Abhishek Singhi, Rich Jones, and Chulwon Lee
Projects & Research
Approximate Random Walk with Restart
Proposed an algorithm computing RWR scores approximately. The proposed algorithm outperforms other state-of-the-art methods in terms of speed and memory efficiency while maintaining high accuracy. I contributed as the first author for a paper accepted in ICDE’18.
Random Walk with Restart on Dynamic Graphs
Proposed an algorithm computing RWR efficiently in time-evolving graphs. The proposed algorithm is the first method that guarantees its exactness on RWR scores on dynamic graphs. The proposed algorithm outperforms previous state-of-the-art dynamic RWR methods. I contributed as the first author for a paper accepted in WWW’18.
Generalized Matrix-Vector Multiplication for Scalable Graph Mining
Proposed an algorithm generalizing matrix-vector multiplication on distributed systems. The proposed algorithm achieves superior scalability than previous graph mining methods. I contributed as co-author for a paper submitted to WSDM’18.
Question Answering System
Proposed Question Answering system which retrieves a correct answer for a given question asked in natural language on any topic. Given several answer candidates, re-ranking module in the system ranks the candidates using Neural Tensor Network. This project is supported by Institute for Information & communications Technology Promotion(IITP) grant funded by the Korea government(MSIP).
SAP HANA Active/Active
Active/Active configuration prepares data in memory for read access in an additional ‘hot standby’ SAP HANA database (a.k.a. the secondary SAP HANA database). Customer could operate two SAP HANA databases. This feature is offered from SAP S/4HANA 1610, FPS01 and SAP HANA 2.0, SPS01. I contributed as co-inventor for two US patents (See Patent above).
SAP HANA Capture and Replay
SAP HANA Capture and Replay captures all incoming SQL statements at the session layer of the SAP HANA database and replays it in the new SAP HANA software to ensure the solution still works correctly.
It reduces testing challenges by recording workload instead of developing it.
SAP HANA Capture and Replay is offered from SAP HANA SPS12.
SAP HANA Session/EAPI layer
Session layer serves as the front-end of SAP HANA database performing thread handling, resource allocation and packet encoding/decoding. EAPI layer is located between Session and SQL layers, and manages connections, statements and transactions which are involved in SQL execution.
Online Motion Puppetry for Non-human Characters
Manipulating characters’ motion whose body structures and motion patters are different from those of human in real time by human’s motion. I contributed as the first author for a paper submitted to KCGS’14.
Crowd Simulation based on Motion Patches
Generating a random crowd simulation in real time using motion data with characters and environments that interact with each other. It generates a random crowd simulation of virtual characters interacting with each other in a non-trivial manner using bunch of raw motion data. I contributed as the first author for a paper submitted to KSPE’13.
AWARDS & HONORS
- National Science & Technology Scholarship, KOSAF (Mar. 2008 - Feb. 2012)
Full tuition exemptions for 8 semesters.
- Cum Laude Graduation Honors, Seoul National University (Feb. 2012)
- C/C++, MATLAB, Python (Advanced) / Java, SQL (Experienced) / HTML (Intermediate)
- Proficient with linear algebra and TensorFlow