When Not to Use ML
summary
The talk examines what machine learning actually is and describes concrete situations where common ML solutions are not justified. It compares ML against traditional analytic methods and explains when to prefer classical analytics over adding ML to a Java project, with speakers Alexander Nozik, Leonid Starikov, Kirill Tolkachev, and Alexey Samosyuk.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 1:21When ML isn't needed
- 4:13Scientific ML approach
- 8:28Engineering over ML
- 11:59Computer vision vs ML
- 18:08ML failures and challenges
- 21:45Tasks without training data
- 25:16AI, ML and CV differences
- 30:18Physics-based models
- 33:31ML applicability limits
- 36:24Working with engineers
- 42:10Training on real data
- 46:27When to use ML
- 47:52Conclusion
related talks
54mML Beyond Metrics: Challenges of Using ML Systems in the Real World
59mMachine Learning: The End of Business Logic?
04mPractical ML Today and Tomorrow
16mMachine Learning--Cybersecurity Boon or Boondoggle
46mRub a Little AI on It
06mMachine learning is about your data and deployment
44mMachine Learning - Opening Pandora's Box
35mML & I