Kitty_Gu - Machine Learning Pathway

Week: 08/04/2020 Update 09/09/20

Overview of Things Learned:

Technical Area:

  • Sparse Gaussian processes (GP) framework - a supervised classification model
  • Active Learning - human in the loop
  • Scrum Workflow
  • Building user interface using Streamlit

Tools:

  • pandas
  • Miro
  • Google Colab
  • Jupyter Notebook
  • GitHub
  • Python
  • Streamlit

Soft Skills:

  • Leandership
  • Perform background research
  • Brainstorm(e.g. for project ideas, potential solutions, team bonding events)
  • Take clear and concise meeting notes
  • Communication
  • Teamwork

Achievement Highlights

  • Researched on tagging recommendation algorithms, and found the MMSG, which is Multi-label Multi-class Sparse Gaussian Process Classification. Also did research on what classifier works the best for a different amount of data and condition.
  • Implemented a web scraper that scrapes posts of the first 5 tags on Stack Exchange.
  • Implemented the very basic manual tagger that takes in a topic title and looks up the corrected tags in the scrapped data.
  • Organized meetings, project parties, and posted movie guessing games for team bonding.
  • Build user interface to incorporate into the active learning loop using Streamlit.

Meetings attended

  • Everyone of them

Goals for the Upcoming Week

  • Start on building the user interface for the annual tagger
  • Incorporate a way to record the low confidence posts in the loop(maybe save it into a csv file)
  • Be more active and communicate with the team more.

Tasks Done

  • Decided on project pipeline -> Active Learning
  • Researched on what algorithm would be best for our situation
  • Implemented a simple web scraper and the very basic of manual tagger