learning-of-the-week

Learning of the Week ^(LOTW)

_{Language is mixed in English and Korean.}
Inspirational learning sources from the week.
The units are in the number of the week, and the number of day (in parentheses) in the year.

🌍 Website: link
🛠️ Github: link —————– 2023

• Week 7
• Week 6
• Week 5
• Week 4
• Week 3
• Week 2

2022

• Week 51
• Week 49
• Week 47
• Week 46
• Week 45
• Week 44
• Week 43

• Week 40

  ---

###

• At Storm Ventures, the mission is collectively made, which is in three words: upbringing startup founders.
  • Storm Ventures CEO Tae Hea Nahm

###

• These days, people use Apache Superset for visualization.
  • SKT
• https://colab.research.google.com/drive/1z31ZfcisCXCSGA5jeid0UNjiHb9oupuV

Week 6

• (43) Difference between Core, Core Plus, Value-Add and Opportunistic Investments. ‘Core’ is synonymous with ‘income’ in the stock market. ‘Core Plus’ is synonymous with ‘growth and income.’ ‘Value-Add’ is synonymous with ‘growth’ in the stock market and is associated with moderate to high risk. Opportunistic is the riskiest of all real estate investment strategies.
  • https://origininvestments.com/2018/02/21/what-are-core-core-plus-value-added-and-opportunistic-investments/

Week 5

• (32) Should know why Selenium agent optioning does not work on certain sites, causing a 403 Error.

Week 2

Week 51

Week 49

• (343) first paint -> first contentful paint -> largest contentful paint. Gzip보다 Brotli로 압축하기. AVIF로 압축하기. 이미지의 불필요한 메타데이터 제거하기(절반까지 용량 절감 가능). 웹폰트 최적화(사용하지 않는 글립에 대한 삭제 및 자체 폰트로 전환). Lighthouse를 사용하기.
  • GDG Frontend 강상진
• (334) There are three kinds of uncertainties. Data (aleatoric uncertainty), model (epistemic uncertainty), and out-of-distribution (novelty detectionn) uncertainty.
  • MODULABS Choi, Sungjoon

Week 47

• (327) During code reviews (in Korea), the roles should be divided as the recorder, the reviewer, the moderator, the reader, and the author. The moderator should lead (or moderate) the meetings for 계획, 킥오프, 수행 및 재작업 단계. The team leader should rather view the process aside.
  • NIPA
• (327) Code review done by certain language-using developers would have difficulty teaching another language-using developer.
  • NIPA
• (326) [Network] TCP 3-way 3rd level handshake. 127.0.0.1 meaning. Port 3306 is MySQL, 23 Telnet. Can you create network system within network layer?
• (326) [Linux] Creating user command line. Simply typing ‘echo’ only displays to terminal, but using ‘»’ with it will not. -f when closing a running application.
• (326) [Container] Namespace isolate include UID, PID, not. User admin access command line.
• (326) [Cloud] S3 names cannot be modified and cannot use underbars and capital letters. Its region cannot be modified as well. Kubernetes services include AKS, EKS, .
• (325)
• (324) The better way to achieve high scores on Quiklab leaderboards is to know how to flexibly use Linux shell commands.
  • Myself
• (324) 앞으로 distributed GPU에 집중할 것이다.
  • Google Next, 박
• (323) 앞으로 마지막으로 명심하면 좋을 것은, 삼성에서도 high level에서 아키텍처를 잘 못해서 후회하고 있다고 한다. Entity별 클래스 등등 잘 분류해서 꼭 칠판에 적어서 설계를 하는 것으로 한다.
  • SPSS-NIPA SW Architecture Seminar
• (323) Test를 잘 쓰려면 적은 test로 최대한 많은 오류를 알아내는 것. 단위 테스트는 statement coverage와 branch coverage로 함.
  • SPSS-NIPA SW Architecture Seminar
• (323) Low code, no code를 잘 하려면 명세서를 잘 써야한다.
  • SPSS-NIPA SW Architecture Seminar
• (323) Reference model (TOGAF, etc.) + architecture pattern (Call-Return, etc.) = reference architecture (J2EE, SOE, etc.)
  • SPSS-NIPA SW Architecture Seminar
• (323) RUP (Rational Unified Project) came before UML. One boundary class per actor/use case pair.
  • SPSS-NIPA SW Architecture Seminar

Week 46

Week 45

(Nov. week 2) of 2022

• (317) Monarcy method curriculum.
  • https://docs.google.com/document/d/18FiJbYn53fTtPmphfdCKT2TMWH-8Y2L-MLqDk-MFV4s/edit
• (316) EfficientNet is a convolutional neural network architecture and scaling method that uniformly scales all dimensions of depth/width/resolution using a compound coefficient.
  • https://paperswithcode.com/method/efficientnet
• (316) In order to decipher edf files (medical data on deep sleep), we can convert them to jpg files and proceed with EfficientNet.
  • https://www.dropbox.com/scl/fi/mgf7tb2e30xlv0dk9qh4i/2.paper?dl=0&rlkey=w7zs340gqvujyswbmo2afie1q

Week 44

(Nov. week 1) of 2022

• (310) 13 portfolios to consider. https://bpowell.co, https://cherupil.com, https://robonilla.com, https://lynnsohn.com, https://chirswelch.co, https://alexlakas.com, https://stijnthijssen.nl, https://yuyangluo.com, https://taamannae.dev, https://yinghsuanho.com, https://karoliskosas.com, https://jaeyouchung.com, https://camillemormal.com
  • https://www.linkedin.com/posts/thijskraan_ux-portfolio-design-activity-6993934856460066816-Ebj2?utm_source=share&utm_medium=member_android
• (309) https://jehyunlee.github.io/2022/11/04/Python-DS-119-kierlecture7/
  • TensorFlow KR

Week 43 ‘22

(Oct. week 5) of 2022

• (300) TensorRT - C++ library
  • https://developer.nvidia.com/tensorrt

Week 40 ‘22

(Oct. week 2) of 2022

• (296) Great (Korean) website for recommender systems: https://root-decimal-c5d.notion.site/Recommender-System-KR-5b773a06e99145e6855bae391c94dc44.
  • Python Data Science Kakaotalk Group
• (291) 리테일과 통신사에서 MSA 전환을 시도하고 있는데, 리테일은 프론트엔드, 통신사는 BSS (Back Server Support)에서 시도한다.
  • MongoDB Day Seoul AWS SA 윤기원
• (291)
• (290)
• (288) Using MAX() would solve the problem of COUNT(sth) * sth, or you could either use SUM().
  • https://stackoverflow.com/questions/29594282/in-sql-how-do-you-multiply-count
• (287) Why is there a ‘ ‘ (space) and AS at the same time in a query? Spaces within [ ] or “” count as a space-recognizable column name.
  • https://stackoverflow.com/questions/49212368/what-is-difference-in-sql-server-between-and
• (287) Does SQL join always need to have anchors?
  • …
• (287) SQL - Difference between UNION, UNION ALL, and JOIN.
  • https://stackoverflow.com/questions/10538539/join-two-select-statement-results
• (286) CPU’s ALU -> GPU’s rendering -> NPU
  • KOREA AI DEVCON 2022 (rebellions_ 박성현 대표)
• (286) ALU (from CPUs) is part of a central processing unit that carries out arithmetic and logic operations on the operands
  • https://www.techtarget.com/whatis/definition/arithmetic-logic-unit-ALU
• (286) IPS,IDS,EDR,NAC - for cloud security.
• (286) ‘MI:DEUM’이란 서비스가 있고, ‘MI:DEUM LETS’이란 플랫폼이 있다.
  • KOREA AI DEVCON 2022 (KT 장두성 상무)
• (286) AI software that writes software software (Jenson Huang, NVIDIA CEO). vGPU: virtual GPU. NVIDIA Omniverse is for developing metaverse.
  • KOREA AI DEVCON 2022 (NVIDIA 정구형 이사)
• (286) ASIC: An application-specific integrated circuit is an integrated circuit (IC) that’s custom-designed for a particular task or application.
  • https://www.arm.com/glossary/asic
• (284) So… k8s is going to cost more, but having a k8s cluster and all of it’s yaml files is going to reduce your vendor lock in. Also, once the cluster is up, managing installs/your apps is going to be easier than managing them in ECS.
  • https://www.reddit.com/r/kubernetes/comments/xlfcs2/what_should_make_me_consider_moving_from_ecs_to/
• (284) If your app is distributed or stateless, and can survive the loss of an instance at any time it’s a great candidate for spot instances.
  • https://www.reddit.com/r/aws/comments/n0mlmb/does_aws_ec2_spot_instance_make_sense_in_a/
• (284) == is for value equality, whereas is is for reference equality.
  • https://towardsdatascience.com/whats-the-difference-between-is-and-in-python-dc26406c85ad
• (284) % is line magic, whereas %% is cell magic in Jupyter.
  • https://stackoverflow.com/questions/58607854/difference-between-and-in-ipython-magic-commands
• (284) ANSI SQL is the abbreviation of American National Standards Institute SQL, which is the standard SQL.
  • https://www.quora.com/What-is-ANSI-SQL-Oracle-SQL-and-MySQL
• (284) 비즈니스 문제가 무엇인지 먼저 잘 판단이 되는 기업이 머신 러닝 솔루션을 배포까지 가장 빠르게 개발한다. wait=False란 파라미터로 Jupyter에서 SageMaker에서 compute하는 것을 기다릴 필요 없이 바로 다음 블록을 진행할 수 있다.
  • AWS Korea AI/ML workshop 이유동 SA
• (284) JupyterLab will eventually replace the classic Jupyter Notebook. Throughout this transition, the same notebook document format will be supported by both the classic Notebook and JupyterLab.
  • https://stackoverflow.com/questions/50982686/what-is-the-difference-between-jupyter-notebook-and-jupyterlab
• (283) There shows a 80-character line when your code gets too long on Colab (there’s also a way to disable it in Preferences).
  • https://stackoverflow.com/questions/56122183/remove-line-at-80-characters-in-google-colab
• (283) ML은 list로 못하는 이유는 병렬처리가 안되기 때문이다. 그래서 matrix를 쓴다.
  • …
• (283) How to do arrows in .md: Up arrow (↑): &uarr; Down arrow (↓): &darr; Left arrow (←): &larr; Right arrow (→): &rarr; Double headed arrow (↔): &harr;
  • https://reimbar.org/dev/arrows/
• (283) Matplotlib: Why is it ‘suptitle’ not ‘subtitle’? → Because it is ‘super-title.’
  • https://stackoverflow.com/questions/59387535/matplotlib-why-is-it-suptitle-not-subtitle
• (283) The distinction between a variable (which denotes a value) and a pointer (which references a value) is subtle and the realm of such advanced topics as programming language semantics, domain theory, and static single assignment; there be dragons.
  • https://www.quora.com/Are-all-variables-pointers
• (282) Python으로 BFS할 때는 시간단순하게 queue.popleft(), DFS는 stack.pop()
  • https://cyc1am3n.github.io/2019/04/26/bfs_dfs_with_python.html