lecture_2022_fall.md

Course Overview, https://github.com/Kwangkee/Gachon/blob/main/Lecture_2022_Fall-1.pdf
Instructor: 이광기 (Kwangkee Lee, [email protected])
시간: 금요일 15:00-18:00

• [1주차에] 본 class 에 기대하는 바, 제안사항 알려주시고, [이후] 질문/제안 언제든 주시기 바랍니다.
• 숙제, 발표자료는 하루 전 까지 [사이버캠퍼스]에 upload 해 주세요.

Slides

• Week-1 Slide: https://github.com/Kwangkee/Gachon/blob/main/slides/FL_Introduction_2022_Fall.pdf
• Week-2 Slide: https://github.com/Kwangkee/Gachon/blob/main/slides/FL_Platforms_2022_Fall.pdf
• Week-4 Slide: https://github.com/Kwangkee/Gachon/blob/main/slides/FL_ClientContribution_2022_Fall.pdf
• Week-7 Slide: https://github.com/Kwangkee/Gachon/blob/main/slides/FL_Medical_2022_Fall.pdf
• FedML, https://github.com/Kwangkee/Gachon/blob/main/slides/FL_Product_FedML_20221027.pdf

Awesome Federated Machine Learning

• https://www.google.com/search?q=awesome+Federated+Learning
• https://github.com/innovation-cat/Awesome-Federated-Machine-Learning
• https://github.com/ChanChiChoi/awesome-Federated-Learning
• https://github.com/chaoyanghe/Awesome-Federated-Learning

---

Week-1

강의 (Instructor)

• [Must-read] Federated Learning Tutorial@NeurIPS 2020, https://sites.google.com/view/fl-tutorial/, [slides] https://drive.google.com/file/d/1QGY2Zytp9XRSu95fX2lCld8DwfEdcHCG/view

  Part I: What is Federated Learning?
  Part II: Federated Optimization
  Part III: Privacy for Federated Learning and Analytics
  Part IV: Open Problems and Other Topics

• FL Intro, https://github.com/Kwangkee/FL/blob/main/[email protected]#fl-introduction

  Communication-Efficient Learning of Deep Networks from Decentralized Data, https://arxiv.org/abs/1602.05629
  Towards Federated Learning at Scale: System Design, https://arxiv.org/abs/1902.01046
  [Recommend] Advances and Open Problems in Federated Learning, https://arxiv.org/abs/1912.04977

• Virginia Smith, https://www.cs.cmu.edu/~smithv/,

  • ML with Large Datasets, Fall 2021, https://10605.github.io/fall2021/

    Introduction (slides, video)
    Federated Learning (slides, video)

  • ML with Large Datasets, Fall 2022, https://10605.github.io/

Week-1 실습 (TA: 김대열 박사)

• [발표] DNN 리류, https://github.com/Kwangkee/Gachon/blob/main/slides/TA_DL_overview.pdf
• [실습] PyTorch/Tensorflow 리뷰/설치, Tutorial/Sample code

---

Week-2

강의 (Instructor)

• [Must-read] FLRA: a reference architecture of federated learning systems, https://github.com/Kwangkee/FL/blob/main/FL%40CSIRO.md#flra-a-reference-architecture-for-federated-learning-systems

• 연합학습 Open Source Platform https://github.com/Kwangkee/FL/blob/main/[email protected]

  [Must-read] Flower: A Friendly Federated Learning Research Framework, https://arxiv.org/abs/2007.14390

• FedScale: Benchmarking Model and System Performance of Federated Learning at Scale, https://github.com/Kwangkee/FL/blob/main/FL%40FedScale.md

  [Must-read] FedScale: Benchmarking Model and System Performance of Federated Learning at Scale, https://arxiv.org/abs/2105.11367
  [Recommend] Swan, https://github.com/Kwangkee/FL/blob/main/FL%40FedScale.md#swan
  [slides] http://www-personal.umich.edu/~fanlai/assets/docs/fedscale-icml-slides.pdf from http://www-personal.umich.edu/~fanlai/
  Open-Source Systems for Federated Learning | Stanford MLSys #48, https://www.youtube.com/watch?v=TcbOMbg4F9g

Week-2 실습 (TA: 김진수)

• [발표] 연합학습 Open Source Platform 리뷰, https://github.com/Kwangkee/Gachon/blob/main/slides/TA_FL%20Open-Source%20Platform.pdf
• [발표] 연합학습 Open Source Platform 리뷰-2, https://github.com/Kwangkee/Gachon/blob/main/slides/TA_FL%20Open-Source%20Platform-2.pdf
• [실습] Flower/FedScale 설치, Tutorial/Sample code

---

Week-3

강의 (Instructor)

• Federated Learning on Non-IID Data Silos: An Experimental Study, https://arxiv.org/abs/2102.02079

  Github: https://github.com/Xtra-Computing/NIID-Bench

Week-3 실습 (TA: 김진수)

• Federated MetaSense, 적응적 연합학습

Week-3 학생 발표

• [양세모] [Must-read] 연합학습시스템에서의 MLOps 구현 방안 연구,
  • http://www.jics.or.kr/digital-library/25791
  • https://github.com/Kwangkee/Gachon/blob/main/slides/%EC%97%B0%ED%95%A9%ED%95%99%EC%8A%B5%20%EC%8B%9C%EC%8A%A4%ED%85%9C%EC%97%90%EC%84%9C%EC%9D%98%20MLOps%20%EA%B5%AC%ED%98%84%20%EB%B0%A9%EC%95%88%20%EC%97%B0%EA%B5%AC.pdf
• [이기훈] Open Source Platform 사용기, 실습 결과, https://github.com/Liky98/Federated_Learning/blob/master/README/Week3%203f7a8157a4b6403ab14d2ee8e5bcf967.md
• [시경요] Open Source Platform 사용기, 실습 결과, https://github.com/qq490800573/FL-Platform-Homework/tree/main/Week3-Homework

Week-4

강의 (Instructor)

• Client Selection, https://github.com/Kwangkee/FL/blob/main/FL%40ClientSelection.md

• [Must-read] Oort: Efficient Federated Learning via Guided Participant Selection, https://github.com/Kwangkee/FL/blob/main/FL%40ClientSelection.md#oort

  • https://www.usenix.org/conference/osdi21/presentation/lai
  • Presentation Video: https://www.youtube.com/watch?v=5npOel4T4Mw
  • slides: https://www.usenix.org/system/files/osdi21_slides_lai.pdf

• FedBalancer, https://github.com/Kwangkee/FL/blob/main/AFL.md#t1-kaist

  • https://nmsl.kaist.ac.kr/projects/fedbalancer/, 이기종 사용자 기기가 포함된 환경에서의 최적화된 사용자 학습 데이터 선택 및 연합학습 라운드의 데드라인 제어를 통한 효율적인 연합학습 알고리즘 연구
  • Slides: https://drive.google.com/file/d/1zlN6er5xOcQgiLQCHaGVGso9r5Mvm_Ms/view
  • Conference Presentation: https://www.youtube.com/watch?v=q3NeIUjSjqg&t=2s

• Etc

  • Ditto: Fair and Robust Federated Learning Through Personalization, https://proceedings.mlr.press/v139/li21h.html
  • YaeJeeCho@CM https://github.com/Kwangkee/FL/blob/main/[email protected]#yae-jee-cho

• FL with Noisy Labels

  • Federated Learning with Noisy Labels, https://arxiv.org/abs/2208.09378
  • Overcoming Noisy and Irrelevant Data in Federated Learning, https://arxiv.org/abs/2001.08300
  • Robust Federated Learning with Noisy Labels, KAIST, https://ieeexplore.ieee.org/document/9713942

• 연합학습 Public Data, PFL Benchmarking, https://github.com/Kwangkee/FL/blob/main/[email protected]

Invited Talk (신재민/KAIST)

• Efficient Federated Learning Approaches on Heterogeneous Clients, 이기종 클라이언트 위 효율적 연합학습 방법론, https://github.com/Kwangkee/Gachon/blob/main/slides/TA_FedBalancer_%EC%8B%A0%EC%9E%AC%EB%AF%BC_20221007.pdf
• FedBalancer, https://github.com/Kwangkee/FL/blob/main/[email protected]#fedbalancer

Week-5

강의 (Instructor)

• 연합학습의 Digital Healthcare 분야 적용 사례
• 디지털 헬스케어를 위한 블록체인 융합 원격임상시험 서비스 개발, https://drive.google.com/drive/u/0/folders/1qV5jN-KspZWYuUZEzx0mzLEYPsomzhko
• Equideum Health, https://github.com/Kwangkee/FL/blob/main/BCFL%40Equideum.md
• rPPG, https://github.com/Kwangkee/rPPG, 원격임상시험 D-2 리뷰

Week-5 실습 (TA: 김대열 박사)

• [발표] rPPG 리뷰 및 실습, https://github.com/Kwangkee/Gachon/blob/main/slides/TA_rPPG_Overview.pdf
• [논문] 딥러닝 기반 rPPG 모델 사용을 위한 경량 모델 연구, https://github.com/Kwangkee/Gachon/blob/5dd38ea670a14d0f0dd71fd789ce879c2be9d9a4/slides/%EC%9D%98%EB%A3%8C%EC%A0%95%EB%B3%B4%ED%95%99%ED%9A%8C_2022%EC%B6%94%EA%B3%84_%EB%94%A5%EB%9F%AC%EB%8B%9D%20%EA%B8%B0%EB%B0%98%20rPPG%20%EB%AA%A8%EB%8D%B8%20%EC%82%AC%EC%9A%A9%EC%9D%84%20%EC%9C%84%ED%95%9C%20%EA%B2%BD%EB%9F%89%20%EB%AA%A8%EB%8D%B8%20%EC%97%B0%EA%B5%AC.pdf
• [논문] Assessment of ROI Selection for Facial Video-Based rPPG(MDPI Sensors 2021): [T3-1] 헬스케어 서비스 분야에서, 원격-PPG 기반 생체징후인식 실세계 적용 시나리오에 적응적 연합학습 기술을 적용하기 위한 기반기술, https://www.mdpi.com/1424-8220/21/23/7923
• [논문] 안면 이미지 데이터를 이용한 실시간 생체 징후 측정 시스템, https://www.kci.go.kr/kciportal/ci/sereArticleSearch/ciSereArtiView.kci?sereArticleSearchBean.artiId=ART002699582

Week-6

강의 (Instructor)

• 연합학습의 medical 분야 적용, https://github.com/Kwangkee/FL/blob/main/[email protected]

  [Must-read] [Contribution-Aware Federated Learning for Smart Healthcare], https://ojs.aaai.org/index.php/AAAI/article/view/21505
  [Recommend] [삼성병원: 신수용 교수]
  Artificial Intelligence in Healthcare, https://biods220.stanford.edu/, Distributed Learning, Security, and Privacy lecture14.pdf, https://biods220.stanford.edu/lectures/lecture14.pdf

---

Week-7

강의 (Instructor)

• 블록체인 융합 연합학습, https://github.com/Kwangkee/FL/blob/main/BCFL%40Korea.md

  [Must-read] 2CP: Decentralized Protocols to Transparently Evaluate Contributivity in Blockchain Federated Learning Environments, https://arxiv.org/abs/2011.07516, Code: https://github.com/cai-harry/2CP
  [Recommend] 연합학습의 인센티브 플랫폼으로써 이더리움 스마트 컨트랙트를 시행하는 경우의 실무적 고려사항, https://github.com/Kwangkee/FL/blob/main/BCFL%40Korea.md#%EB%B6%80%EA%B2%BD%EB%8C%80

• [Recommend][한글 유튜브] Web3 기술 인프라 , https://youtu.be/2cpUO1XN528?t=3173

• [Udemy Course] https://www.udemy.com/course/master-ethereum-and-solidity-programming-with-real-world-apps/?utm_source=email-sendgrid&utm_medium=1771964&utm_campaign=2022-10-20&utm_term=50683248&utm_content=promo

실습 (TA: 고은수/이노피아)

• [발표, 실습] BCFL, https://github.com/Kwangkee/Gachon/blob/main/slides/TA_BCFL_%EA%B3%A0%EC%9D%80%EC%88%98.pdf

---

Week-8

강의 (Instructor)

• 블록체인 융합 연합학습, https://github.com/Kwangkee/FL/blob/main/BCFL%40Korea.md

  [Must-read] [Towards Trustworthy AI: Blockchain-based Architecture Design for Accountability and Fairness of Federated Learning Systems], https://scholar.google.com/citations?view_op=view_citation&hl=ko&user=TuL21poAAAAJ&sortby=pubdate&citation_for_view=TuL21poAAAAJ:koF6b02d8EEC [Recommend] https://github.com/Kwangkee/FL/blob/main/BCFL%40Korea.md#%EC%84%9C%EC%9A%B8%EB%8C%80

---

Week-9

강의 (Instructor)

• Client Contribution, https://github.com/Kwangkee/Gachon/blob/main/slides/FL_ClientContribution_2022_Fall.pdf

Invited Talk (성균관대 신성국/김광수 교수님)

• Client Contribution, https://github.com/Kwangkee/Gachon/blob/main/slides/InvitedTalk_Client_Contribution_%EC%8B%A0%EC%84%B1%EA%B5%AD_%EC%84%B1%EA%B7%A0%EA%B4%80%EB%8C%80_20221028.pdf

Paper Review

• 양세모, Contribution-Aware Federated Learning for Smart Healthcare, https://giai.notion.site/Contribution-Aware-Federated-Learning-for-Smart-Healthcare-19a363e1b60e4427b297f58dafdb1833
• 김진수, Where to Begin? Exploring the Impact of Pre-Training and Initialization in Federated Learning, https://github.com/Kwangkee/Gachon/blob/main/slides/Paper_Pretrained_%EA%B9%80%EC%A7%84%EC%88%98_20221028.pdf

---

Week-10

강의 (Instructor)

• FL Product, https://github.com/Kwangkee/Gachon/blob/main/slides/FL_Product_FedML_20221104.pdf

• FedML: https://fedml.ai/

  • FedML Product Overview, https://medium.com/@FedML/fedml-ai-platform-releases-the-worlds-federated-learning-open-platform-on-public-cloud-with-an-8024e68a70b6
  • Tutorial for Platform, https://fedml.ai/platform-tutorial/

• FedNLP: A Research Platform for Federated Learning in Natural Language Processing, https://github.com/FedML-AI/FedML/tree/master/python/app/fednlp

Invited Talk (성균관대 우홍욱 교수님)

• Reinforcement Learning with Pretrained Models, https://github.com/Kwangkee/Gachon/blob/main/slides/20221104_PretrainedRL_CSSeminar_%EC%9A%B0%ED%99%8D%EC%9A%B1%EA%B5%90%EC%88%98.pdf

---

Week-11

Invited Talk (인하대 이선우 교수님)

• https://scholar.google.com/citations?hl=en&user=WA9KNNcAAAAJ&view_op=list_works&sortby=pubdate
• Scalabl Federated Learning on Real World Edge Device Environments, https://github.com/Kwangkee/Gachon/blob/main/slides/Scalable%20FL%20%EC%9D%B4%EC%84%A0%EC%9A%B0%20%EA%B5%90%EC%88%98%EB%8B%98_2022_Fall.pdf

---

Week-12

강의 (Instructor)

• 동적인 디바이스 환경에서 적응적 연합학습, https://github.com/Kwangkee/FL/blob/main/AFL.md
• Personalized federated learning, https://github.com/Kwangkee/FL/blob/main/[email protected]#towards-personalized-federated-learning

  [Must-read] Towards Personalized Federated Learning, https://arxiv.org/abs/2103.00710

김진수

• 원격 PPG에 대한 메타러닝 기반 개인화 연합학습 적용 및 성능 평가
• 발표자료, https://github.com/Kwangkee/Gachon/blob/main/slides/%EB%8C%80%ED%95%9C%EC%9D%98%EB%A3%8C%EC%A0%95%EB%B3%B4%ED%95%99%ED%9A%8C%EB%B0%9C%ED%91%9C%EC%9E%90%EB%A3%8C_%EB%A9%94%ED%83%80%EB%9F%AC%EB%8B%9D%20%EA%B8%B0%EB%B0%98%20%EA%B0%9C%EC%9D%B8%ED%99%94%20%EC%97%B0%ED%95%A9%ED%95%99%EC%8A%B5_%EA%B9%80%EC%A7%84%EC%88%98.pdf
• 논문, https://github.com/Kwangkee/Gachon/blob/main/slides/%EC%9D%98%EB%A3%8C%EC%A0%95%EB%B3%B4%ED%95%99%ED%9A%8C_2022%EC%B6%94%EA%B3%84_%EC%9B%90%EA%B2%A9%20PPG%EC%97%90%20%EB%8C%80%ED%95%9C%20%EB%A9%94%ED%83%80%EB%9F%AC%EB%8B%9D%20%EA%B8%B0%EB%B0%98%20%EA%B0%9C%EC%9D%B8%ED%99%94%20%EC%97%B0%ED%95%A9%ED%95%99%EC%8A%B5%20%EC%A0%81%EC%9A%A9%20%EB%B0%8F%20%EC%84%B1%EB%8A%A5%20%ED%8F%89%EA%B0%80.pdf

양세모

• Universal EHR Federated Learning Framework, https://giai.notion.site/Universal-EHR-Federated-Learning-Framework-db4f84effac74dd7838b3f8d80d23ef9
• FedML Android Client, https://giai.notion.site/FedML-9b5df070516a43fa8c01882db529b13e

---

Week-13

강의 (Instructor)

Invited Talk (서울과학기술대학교 컴퓨터공학과 박종혁 교수님)

• Secure Electronic Health Record Sharing Scheme based on Federated Learning in Medical Informatics, https://ieeexplore.ieee.org/abstract/document/9774951

TA (고은수/이노피아)

• 블록체인 기반 연합학습을 위한 레퍼런스 아키텍처, https://github.com/Kwangkee/Gachon/blob/main/samples/%EB%B8%94%EB%A1%9D%EC%B2%B4%EC%9D%B8%20%EA%B8%B0%EB%B0%98%20%EC%97%B0%ED%95%A9%ED%95%99%EC%8A%B5%EC%9D%84%20%EC%9C%84%ED%95%9C%20%EB%A0%88%ED%8D%BC%EB%9F%B0%EC%8A%A4%20%EC%95%84%ED%82%A4%ED%85%8D%EC%B2%98_%EB%B0%A9%EC%86%A1%EB%AF%B8%EB%94%94%EC%96%B4%EA%B3%B5%ED%95%99%ED%9A%8C_20221006.pdf
• 블록체인 기반 연합학습을 위한 레퍼런스 아키텍처, https://github.com/Kwangkee/Gachon/blob/main/samples/%EB%B8%94%EB%A1%9D%EC%B2%B4%EC%9D%B8%EC%97%B0%ED%95%A9%ED%95%99%EC%8A%B5_%EC%9D%98%EB%A3%8C%EC%9D%B8%EA%B3%B5%EC%A7%80%EB%8A%A5%ED%95%99%ED%9A%8C_2022%EC%B6%94%EA%B3%84_20221013.pdf

---

TA

주차	TA 발표	실습	담당 TA
1	DNN 리뷰, PyTorch/TF 리뷰, DNN 환경	Git 계정, PyTorch 설치, Tutorial/Sample code	김대열
2	연합학습 Open Source Platform (Flower/FedScale) 리뷰	설치, Tutorial/Sample code	김진수
3	Federated MetaSense, 적응적 연합학습	적응적 연합학습	김진수
4	연합학습 Public Data, Benchmarking 리뷰	PFL Benchmarking 설치, Tutorial/Sample code	김진수
5	rPPG 리뷰	rPPG code 리뷰	김대열
6	원격임상시험 D-3 리뷰	PoC 결과	양세모
7	BC 리뷰	Ethereum Simulator, Solidity	고은수
8	BCFL 리뷰	2CP Simulator	고은수

---

[etc]

• [Invited Talk] KAIST MetaSense, FedBalancer
• Stanford MLSys Seminar Series, https://mlsys.stanford.edu/
• MLOps

  MLOps System Design for Development and Production | Stanford MLSys #44, https://www.youtube.com/watch?v=TcbOMbg4F9g

• Virginia Smith, https://www.cs.cmu.edu/~smithv/, ML with Large Datasets, Fall 2021, https://10605.github.io/fall2021/

  Aug 31 Introduction (slides, video)
  Nov 23 Federated Learning (slides, video)

• CS 330: Deep Multi-Task and Meta Learning, http://cs330.stanford.edu/
• BIODS220 (CS271, BIOMEDIN220) Artificial Intelligence in Healthcare, https://biods220.stanford.edu/

  Lecture 14 Nov 15 (Mon) Distributed Learning, Security, and Privacy lecture14.pdf, https://biods220.stanford.edu/lectures/lecture14.pdf