Questions and Insights

Table of Contents

1. Machine Learning Questions

  • What are the factors in selecting new features
  • Selecting models
  • How does one test new features
  • How does one produce new features
  • How does one deploy new features
  • What is the process for regression
  • What visualization can be provided

http://www.scikit-yb.org/en/latest/tutorial.html https://docs.aws.amazon.com/machine-learning/latest/dg/ml-model-insights.html

https://aws.amazon.com/aml/faqs/

https://medium.freecodecamp.org/a-beginners-guide-to-training-and-deploying-machine-learning-models-using-python-48a313502e5a https://engineering.quora.com/Avoiding-Complexity-of-Machine-Learning-Systems

https://archive.ics.uci.edu/ml/datasets/Wine+Qualityhttps://archive.ics.uci.edu/ml/datasets/Wine+Quality

These are some of the areas we would want to consider (as well as areas that we don't want to consider):

2. Reading

3. Nation States Sponsors

4. Current Uses

4.7. Reinforcement Learning: Example

https://becominghuman.ai/reinforcement-learning-step-by-step-17cde7dbc56c https://github.com/rfeinman/tictactoe-reinforcement-learning

  • Create Environment (agent_state, agent_move)
  • Create State
  • Specify reward
  • Create game.py https://inventwithpython.com/chapter10.html
  • Create agent.py
  • Create Actions list
  • Create game visualization
  • Create match visualizations

https://skymind.ai/wiki/deep-reinforcement-learning#code

This would also look at game boards.

https://github.com/topics/tic-tac-toe?l=javascript&o=desc&s=stars

4.17. AdTech

https://www.spotx.tv/resources/blog/product-pulse/artificial-intelligence-vs-machine-learning-disrupting-ad-tech/ https://www.teads.tv/machine-learning-teads-4-use-cases-adtech-industry/

View-through rate prediction (VTR) Broken creative detection Bid-request relevancy prediction Look-alike Modeling (on-going)

4.18. FinTech

https://rubygarage.org/blog/machine-learning-in-fintechFraud prevention Risk management Investment predictions Customer service Digital assistants Marketing Network security Loan underwriting Algorithmic trading Process automation Document interpretation Content creation Trade settlements Money-laundering prevention Custom machine learning solutions

4.30. Training

5. Approaches

  • What is Supervised Learning?
  • What is Unsupervised Learning?
  • What is Reinforcement Learning?

6. MLaaS vs. Hosted

7. Machine Learning Use Cases

8. Product Areas

9. Hiring and Skillsets

Focus on executives, designers, and business analysts.

10. Summarizing Entities

11. Creating Data Pipelines

12. Cleaning Data

13. Finding Features

14. Evaluating Features

15. Feature Importance and Correlated Features

16. Evaluating Models

17. Maintaining Models

18. DataLab and Experimentation

19. DataLab vs. Production

20. Outlier Management

21. Tracking Technical Debt

22. Design and Code Review

23. Documentation

24. Outliers   technical_debt

25. Monitoring Correlated Features

26. Causation   technical_debt

27. Boundary Conditions   technical_debt

One sense in which all machine learning algorithms incur a technical debt is through the erosion of boundaries

https://www.kdnuggets.com/2015/01/high-cost-machine-learning-technical-debt.html

28. Monitoring Outcomes

29. Escalation

30. Scenarios