Tushar_432 - Machine Learning Pathway

Certificates June 2020 Certificates Machine Learning Pipeline (NLP) Self-Assessments
1

Overview of thing learned
Technical: I learned web scraping using the selenium package for python. I used it for scraping the content of Amazon sellers’ discourse-based website.
Tools: I learned using slack, GSuite emails for professional works, and got to use GitHub for collaborating on a project.
Soft Skills: I got to work on a team project which helped me learn teamwork, discussing ideas within members, effective communication.

Three Achievement highlights

  1. Learned and successfully deployed web scraper for the first time.
  2. Collaborated with team to identify problems, solve them, and implement.
  3. Could deliver the required data in the expected format with the help of the team.

List of meetings/training attended
Meetings:

  1. Team Introduction
  2. Accounts set-up and Project overview
  3. Selenium demo and tasks assignment
  4. Progress discussion

Goals for upcoming weeks
Learn and use BERT.
Use the data extracted

Tasks Done

  1. Task: Set up mentorchains, Asana, and Slack accounts.
    Hurdles: A bit of confusion in the process.
    I could solve it with the help of team leads.
  2. Task: Learn and get familiar with Selenium and discourse forum (Amazon sellers).
    Hurdles: New to Selenium so couldn’t get much.
    Technical lead helped with resources to learn from and sample code.
  3. Task: Extract and format data from the Amazon sellers forum.
    Hurdles: Some bugs in code.
    Fixed all the errors with the help of team discussion and technical lead.
1 Like
2

Overview of thing learned
Technical: I learned the concepts of Natural Language Understanding, transformers, and attentions.
Tools: I learned using distilBERT for NLU.
Soft Skills: I got to work on a team project which helped me learn teamwork, discussing ideas within members, effective communication.

Achievement highlights

  1. Successfully delivered the required data.
  2. Learned BERT and concepts of NLU for first time.

List of meetings/training attended
Meetings :

  1. Discussion of BERT
  2. distilBERT Modeling

Goals for upcoming weeks
Apply distilBERT on data and perform classification.

Tasks Done

  1. Task : Understand the concepts behind BERT and try implementation.
    Hurdles : Unfamilier with topic.
    I could comprehend it after reading some articles online.
3

Overview of thing learned
Technical: I got to learn the implementation of distilBERT. Then I used distilBERT for training text for posts classification.
Tools: I learned to use the transformers package of python.
Soft Skills: I got to work on a team project which helped me learn teamwork, discussing ideas within members, effective communication.

Achievement highlights

  1. Successfully implemented distilBERT for post classification.
  2. Achieved accuracy of 83.5% for classification.
  3. Got to know and worked with some awesome people.

List of meetings/training attended
Meetings :

  1. distilBERT modeling.
  2. Subteam presentations.

Goals for upcoming weeks

  1. Continue for 3 more weeks.
  2. Learn React.
  3. Improve model.
  4. Deploy model using AWS.

Tasks Done

  1. Task: Preprocess data
    Hurdles: The model was giving errors.
    Checked the data and found some null values. Dropped those NaN values and then model worked well.
  2. Task : Implemented distilBERT.
    Hurdles: Colab crashing because of the large size of the input to tensor.
    Had to reduce the size of tokens to 150.
  3. Task: Classified trained data and tried various classifiers.
    Hurdles: None