Jijoong Moon

Samsung Research, Samsung Software Engineer - Seoul, Korea

Dec 2019-Current

NNtrainer is an Open Source Project. The aim of the NNtrainer is to develop Software Framework to train neural network model on embedded devices which has relatively limited resources. Rather than training the whole layers, NNtrainer trains only one or a few layers added after the feature extractor. It is available at https://github.com/nnstreamer/nntrainer. I am maintainer and founder of this project.

June 2018-Current

NNStreamer is a set of Gstreamer plugins, which allows Gstreamer developers to adopt neural network models easily. It is an opensource project and downloaded from https://github.com/nnstreamer/nnstreamer. I have been a main contributor since the beginning of this project.

May-Jan 2018

Developing Tizen Platform for the Autonomous Driving which includes various soft ware packages such as ROS(Robot OS System) and Neural Network Frameworks(Caffe,tensorflow, etc.). Used Tizen Build System including OBS(Open Build Service) and package management system. Successfully finished auto driving test on top of this platform.

Mar-Dec 2017

NPU Compiler generates binary to execute inference on NPU HW. I developed fron tends(parser, optimizer) mainly, which takes neural network framework model, parses and generates binary for each neural network layers. It also includes opti mizatino for better performance such as removing memory transfer overhead, blocking technique, etc.. Successfully executed googlenet and other neural network model

Oct 2016-Feb 2017

CheckedC( https://github.com/Microsoft/checkedc) is additional C/C++ language specification to make robust against security attack such as buffer overflow. It was joint project with Microsoft Research and I was main contributor at Samsung. Most of work was implementation of language specification using LLVM/CLANG, which was improved through discussion.

Mobile Division, Samsung Software Engineer - Seoul, Korea

Oct 2016-Feb 2017

I was involved security team and had developed on-device encryption and decryption system software. It works across whole process from start and I developed system SW to manage keys for block devices.

Mar-Sep 2015

Analyzed Software including Kernel and boot sequence. Trying to detect potential vulnerabilities of Mobile Device. Wrote PoC code for rooting and demonstrated security attack. Also applied security patches about known vulnerabilities such as CWE.

DMC R&D Center, Samsung Software Engineer - Seoul, Korea

Jul 2014-Feb 2015

Had Developed AOT Compiler Technique to convert android application to Tizen Application. Also, looked for way to integrate auto parallelization technique for multicore into ART (Android Runtime) using LLVM/CLANG. It was pure research project to evaluate the possibility.

Jan 2013-June 2014

I developed auto parallelization technique based on Alias Analysis for Multi-Core with LLVM for Embedded Device. OpenMP(Open Multi Procssing) framework which support 4.0 Specification was improved for low power computing with LLVM/CLANG.

Aug 2010-May 2011

I optimized android application with NEON SIMD and reseach about GPGPU Computing for embedded device. Developed high performance Image conversion library.

College of Engineering, SNU Software Engineer - Seoul, Korea

Jul 2008-Jul 2010

I was a maintainer in college of engineering during my study. Developed and Maintained the web service (included developing HA-High Availability cluster system) and HPC (High Performance Computing) cluster for scientific computation. Tool chain service was included such ad ABAQUS & LS-DYNA. Also I gave lecture about how-to Build cluster includes diskless.

Seoul National University Seoul, Korea

July 2010

Major: Aerospace Engineering
Thesis: "High Velocity Contact-Impact Simulation with Parallel Computing"
Adviser: Prof. SeungJo Kim

Feb 2004

Major: Aerospace Engineering
Thesis: "Massive Parallel Computing for Finite Element Method"
Adviser: Prof. SeungJo Kim

Research Interests

• 520 CPUs, 1.24 TFlops : Top500 56th in 2003
• It was the first cluster system achieved over 1TFlops in Korea.
• Accelerate Performance using MPI, OpenMP, OpenCL, GPGPU Computing

• Developed Parallel Lagrangian-Eulerian Hydrocode with MPI
• Numerical Library Acceleration : BLAS, ATLAS, LINPACK, LAPACK, etc.
• Develop High Performance Direct/Iterative Solver

• Developed Departmental Computing GRID System for Structural Analysis

Peer-Reviewed journal Papers

1. Jijoong Moon, SeungJo Kim, and Minhyung Lee. High velocity impact simulation of brittle materials with node separation scheme in parallel computing environment. Computer Modeling in Engineering & Sciences , 59(3):275–300, 2010
2. Seung Hoon Paik, Jijoong Moon and SeungJo Kim. Parallel performance of large scale impact simulation on linux cluster super computer. Computers & Structures , 84(10-11):732–741, 2006

Peer-Reviewed Conference Papers

1. Jijoong Moon, Hyun Suk Lee, Jiho Chu, Donghak Park, Seungbaek Hong, Hyungjun Seo, Donghyeon Jeong, Sungsik Kong, MyungJoo Ham A New Frontier of AI: On-device AI Training and Personalization 46th ACM/IEEE International Conference on Software Engineering (ICSE)-SEIP, Lisbon, Portugal, April, 2024
2. Francesco Barbato, Umberto Michieli, Jijoong Moon, Pietro Zanuttigh, Mete Ozay Cross-Architecture Auxiliary Feature Space Translation for Efficient Few-Shot Personalized Object Detection The 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems, ADNEC, Abu Dhabi, UAE, October, 2024
3. Kirill Paramonov, Jia-Xing Zhong, Umberto Michieli, Jijoong Moon, Mete Ozay Swiss DINO: Efficient and Versatile Vision Framework for On-device Personal Object Search The 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems, ADNEC, Abu Dhabi, UAE, October, 2024
4. Elena Camuffo, Umberto Michieli, Simone Milani, Jijoong Moon, Mete Ozay Enhanced Model Robustness to Input Corruptions by Per-corruption Adaptation of Normalization Statistics The 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems, ADNEC, Abu Dhabi, UAE, October, 2024
5. Hafiz Tiomoko Ali, Umberto Michieli, Jijoong Moon , Daehyun Kim, Mete Ozay Deep Neural Network Models Trained with a Fixed Random Classifier Transfer Better Across Domains 2024 IEEE International Conference On Acustics, Speech and Signal Processing, Seoul, Korea, April, 2024
6. Elena Camuffo, Umberto Michieli, Jijoong Moon, Daehyun Kim, Mete Ozay FFT-Based Selection and Optimization of Statistics for Robust Recognition of Severely Corrupted Images. 2024 IEEE International Conference On Acustics, Speech and Signal Processing, Seoul, Korea, April, 2024
7. Umberto Michieli, Jijoong Moon , Daehyun Kim, Mete Ozay Object-Conditioned Bag of Instances For Few-shot Personalized Instance Recognition 2024 IEEE International Conference On Acustics, Speech and Signal Processing, Seoul, Korea, April, 2024
8. MyungJoo Ham, Jijoong Moon , Geunsik Lim, Jaeyun Jung, Hyoungjoo Ahn, Wook song, Sangjung Woo, Parichay Kapoor, Dongju Chae, Gichan Jang, Yongjoo Ahn, Jihoon Lee Nnstreamer: Efficient and Agile Development of On-Device AI System 43rd ACM/IEEE International Conference on Software Engineering (ICSE)-SEIP, Madrid, Spain, May, 2021
9. Geunsik Lim, MyungJoo Ham, Jijoong Moon, Wook song LightSys: Lightweight and Efficient CI System for Improving Integration Speed of Software 43rd ACM/IEEE International Conference on Software Engineering (ICSE)-SEIP, Madrid, Spain, May, 2021
10. Jijoong Moon , MyungJoo Ham, Geunsik Lim, Performace Analysis of Neural Network Pipelining for Multimodal On-Device AI Application. Korea Software Conference, PyungChang, Korea, Dec. 2019
11. Geunsik Lim, MyungJoo Ham, Jijoong Moon , Wook song, Sangjung Woo, and Sewon Oh. Taos-ci: Lightweight & modular continuous integraion system for edge computing. In IEEE International Conference on Consumer Electronics , 2019
12. Jijoong Moon , SeungJo Kim, Minhyung Lee, and Yohan Yoo. Optimization for parallel impact analysis of brittle materials using dynamic load balancing. In 1st Conference on Agency for Defense Development , 2009
13. Jijoong Moon , SeungJo Kim, Minhyung Lee, and Yohan Yoo. Parallelization technique for high-velocity contact/impact analysis using dynamic load balancing. In The Korea Institute of Military Science and Technology , 2009
14. Casper Adriaan Buizer, Jijoong Moon , and SeungJo Kim. Low-velocity impact analysis of composite material using buffer zone interface of micro/macro modeling. In Conference on Korean Society for Composite Materials , 2009
15. Jijoong Moon , SeungJo Kim, and Minhyung Lee. High-speed contact/impact simulation code. In 8th World Congress on Computational Mechanics , 2008
16. Jijoong Moon , SeungJo Kim, and Minhyung Lee. Parallel strategies of net3d for high- speed impact simulation. In International Conference on Computational & Experimental Engineering and Sciences , 2008
17. Jijoong Moon , SeungJo Kim, and SeungHoon Paik. Parallel performance of ipsap/explicit for contact-impact simulation. In 9th US National Congress on Computational Mechanics , 2007
18. Y. Yoon, J. Moon , E. Lee, S. Kim, and J. Kim. Parallel optimal design of satellite bus struc- tures using particle swam optimization. In 48th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics & Materials Conference , 2007
19. Minhyung Lee, Seung Hoon Paik, Jijoong Moon , and SeungJo Kim. Parallel compu- tation of two-step eulerian scheme on multi-node supercomputers. In International Conference on Computational & Experimental Engineering and Sciences , 2007
20. Min Ki Kim, Jijoong Moon , and SeungJo Kim. Wccs: Windows comput cluster server for hpc and its performance & applications. In KSIAM Annual Conference , 2007
21. Jijoong Moon , and SeungJo Kim. Parallelization and performance evaluation of contact/impact simulation. In KSIAM Spring Conference , 2007
22. Jin Woo Park, Si Hyoung Park, Jijoong Moon , Youngha Yoon, and Seung Jo Kim. High- fidelity simulation based optimum design utilizing computing grid technology. In 46th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics & Materials Conference , 2006
23. Jin Woo Park, Si Hyoung Park, In Seong Hwang, Jijoong Moon , Youngha Yoon, and Seung Jo Kim. Optimal blade system design of a new concept vtol vehicle using the departmental computing grid system. In Super Computing Conference 2004 , 2004

1. Jijoogn Moon, Myung Joo Ham, Geunsik Lim, Wook Song, sang-Jung Woo, Jaeyun Jung SECURITY ENHANCEMENT OF NEURAL NETWORK MODELS 2020
2. Myung Joo HamJijoogn Moon, Jin Hyuck Park, Hyoung Joo Ahn, Geunsik Lim, Jaeyun Jung ELECTRONIC APPARATUS FOR CONTROLLING DATA PROCESSING OF MODULED NEURAL NETWORK AND THEREOF CONTROL METHOD, 2018
3. Jijoogn Moon, JeonCheol-Yong, KimRakie, JangYongseok TECHNIQUE FOR DYNAMICALLY CONTROLLING PROCESSING DEVICES IN ACCORDANCE WITH CHARACTERISTIC OF USER APPLICATION, 2015
4. Uison Yoon, KimRakie, Jijoogn Moon, METHOD AND APPARATUS FOR EXECUTING SOFTWARE IN A ELECTRONIC DEVICE, 2014

Languages
C/C++, Python, Java, PHP, Shell Programming, MPI, OpenMP, OpenACC, OpenCL, CUDA, etc.

Tools
LLVM/CLANG, Gstreamer, emacs, Build Tools (CMake, Make, meson, etc.) MatLab, LINUX, Git, etc.

OpenSources
https://github.com/nnstreamer/nntrainer
https://github.com/nnstreamer/nnstreamer
https://github.com/nnstreamer/TAOS-CI
https://github.com/Microsoft/checkedc