Research : Machine-Learning based Network (머신러닝 기반 네트워크 연구)

Research Area (연구분야: 머신러닝 기반 네트워크)

연구실 소개

모바일컴퓨팅연구실 소개.pdf

Recent Publications

  • 사전학습 BERT 기반 미세 조정 기법별 감정 분류 성능 평가
    • 제3회 인공지능학술대회, 2022년 9월28일(수)~30일(금) 
    • 홍성규, 김상철
    • BERT(Bidirectional Encoder Representations from Transformers)는 2017년 구글에서 발표한 Transformer 기반의 양방향 문맥을 학습할 수 있도록 고안된 언어모델이다. 오늘날 언어모델의 규모가 커짐에 따라, 현실적으로 리소스 및 시간 등 학습하는데에 어려움이 수반하고 있다. 이러한 문제를 해결하기 위해 사전학습된 언어를 Fine-tuning 하는 과정에서 파라미터를 경량화 시키기 위해 대표적으로 AdapterFusion과 LoRA 와 같은 Fully fine-tuning 을 대체하는 방법이 연구되고 있다. 본 논문에서는 새로운 Fine-tuning 기법들을 BERT 에 적용하며, 학습 시간 및 정확도에 어느정도 영향을 끼치는지 연구한다.
  • Survey of Optimization Algorithms
    • SWCC2022, 2022-08-23
    • Mouna Lamine, Sang-Chul Kim
    • Machine Learning is facing a rapid development due to the nearly limitless quantity of available data, and it is widely applied in various fields. Optimization is one of the core components in the machine learning which attracted more researcher’s attention to it. During the past few years, a lot of work on improving the optimization methods in machine learning have been proposed successively. In this paper we will introduce the different Optimization first order algorithms, their mathematical representation and their source code.
  • 공간 혼잡도 산출 장치 및 방법, 이를 기록한 기록매체
    • APPARATUS AND METHOD OF CALCULATING SPATIAL CONGESTION, STORAGE MEDIA STORING THE SAME)
    • 김상철, 2020-02-07
  • Human Activity Recognition and Prediction Based on Wi-Fi Channel State Information and Machine Learning
    • 2019 International Conference on Artificial Intelligence in Information and Communication (ICAIIC), Feb., 2019
    • Sang-Chul Kim, Tae Gi Kim, Sung Hyun Kim
  • A Study of CNN-Based Human Behavior Recognition with Channel State Information
    • 35th International Conference on Information Networking (ICOIN), Jan. 13~16, 2021
    • K. Hwang and Sang-Chul Kim,
    • Slides, Video_talk 
  • Device-free Activity Recognition Using CSI & Big Data Analysis : A Survey
    • 2017 Ninth International Conference on Ubiquitous and Future Networks (ICUFN), July. 2017
    • Sang-Chul Kim

Recently Invited Tutorial

  • 파이썬 모델링 프로그래밍, 김상철
    • 머신러닝/인공지능 기초/응용 SW 프로그래밍 강좌, 
    • 한국통신학회, 7월 28일~30일, 2021, https://event.kics.or.kr/594/program  
    • Title: 파이썬 모델링 프로그래밍 
    • Abstract: 본 강좌에서는 파이썬을 이용한 데이터 분석 및 시각화, 1차원입력직선모델링, 2차원입력평면모델링,선형 기저 함수 모델링 알고리즘 등의 머신러닝 기초를 학습하고 프로그래밍으로 실습한다.
  • Python-based Machine Learning Algorithmic Thinking
    • Sang-Chul Kim 
    • 1st Korea Artificial Intelligent Conference,  Dec.16~18, 2020,
    • https://www.manuscriptlink.com/society/kics/conference/koreaai/programBook 
    • Title: Python-based Machine Learning Algorithmic Thinking, slide
    • Abstract: In this tutorial, the installation procedure of Pycham, Anaconda, and TensorFlow for machine learning will be introduced and debugging technology using Pycharm IDE will be addressed through data visualization examples. Numerical algorithms for machine learning are explained and programmed by the Pycharm debugging process. After looking at the strategy of Curve-Fitting, we look at supervised learning of machine learning through TensorFlow debugging. We will understand the algorithm of gradient descent for analyzing the loss function and explain multivariable regression analysis using the scikit-learn module.
    • 이 튜토리얼에서는 머신 러닝을 위한 Pycham, Anaconda, TensorFlow의 설치 절차를 소개하고 데이터 시각화 예제를 통해 Pycharm IDE를 사용한 디버깅 기술을 다룹니다. 기계 학습을 위한 수치 알고리즘은 Pycharm 디버깅 프로세스에 의해 설명되고 프로그래밍됩니다. Curve-Fitting의 전략을 살펴본 후 TensorFlow 디버깅을 통한 기계 학습(지도 학습)을 살펴 봅니다. 손실 함수 분석을 위한 경사 하강법 알고리즘을 이해하고 scikit-learn 모듈을 이용한 다변수 회귀 분석을 설명합니다.

Selected Journal Publications


  1. Driving Mode Switching Mechanism based on Reinforcement Learning 
    1. Yilin Zhu  and Sang-Chul Kim
    2. Journal of Theoretical and Applied Information Technology, 2023
  2. SALA: Smartphone-Assisted Localization Algorithm for Positioning Indoor IoT Devices 
    1. J. Jeong, S. Yeon, T. Kim, H. Lee, and Sang-Chul Kim
    2. Wireless Networks, 2018
  3. Development of maze escape algorithm using Raspberry Pi-based moving object for the services of unmanned automobile
    1. Sang-Chul Kim and Goo-Hoon Byun
    2. Journal of Theoretical and Applied Information Technology, 2018
  4. Performance Comparison of MISP-based MANET Strong DAD Protocol
    1. Sang-Chul Kim
    2. Transactions on Internet and Information Systems, 2015
  5. An Entropy-based Stability Algorithm for Regulating the Movement of MANET Nodes
    1.  Sang-Chul Kim,
    2. Transactions on Internet and Information Systems, 2011
  6. Modeling and Matching Design of a Tension Controller Using Pendulum Dancer in Roll-to-Roll Systems
    1. H. Kang, C. Lee, K. Shin, and Sang-Chul Kim
    2. IEEE Transactions on Industry Applications, 2011
  7. Minimized Power Consuming Adaptive Scheduling Mechanism for Cluster-based Mobile Wireless Networks
    1. J. -M. Chung, Sang-Chul Kim, W.-C. Jeong, and S.-S. Joo
    2. IET Electronics Letters, 2009
  8. Analysis of Link Error Effects in MANET Address Autoconfiguration Protocols
    1. Sang-Chul Kim and J.-M. Chung
    2. Journal of Communications and Networks, 2009
  9. Message Complexity Analysis of Mobile Ad Hoc Network Address Autoconfiguration Protocols
    1.  Sang-Chul Kim, J.-M. Chung
    2.  IEEE Transactions on Mobile Computing, 2008
  10. Performance Analysis of Entropy-based Multi-Robot Cooperative Systems in a MANET
    1. Sang-Chul Kim, K. Shin, C. Woo, Y. Eom, and J. Lee
    2. International Journal of Control, Automation, and Systems, 2008 
  11. A Scalability Comparison of GST and SST MANET Multicast Routing Algorithms
    1. Sang-Chul Kim
    2. Elsevier Computer Communications, 2007