Learning about machine learning
Using the teach skill recently to learn more about the math behind machine learning.
While youtube videos, articles, books, all do a great job, I find myself wanting more in terms of actually internalising the concepts. Working with an LLM works wonders as I can get it to validate my weird mental models (to make sure that its right) so it actually sticks with me.
On top of that, it helps that I can actually track my progress as my openclaw can remember notes about my learnings (since it has access to a persistent filesystem), and how I actually want to learn. This makes it much more engaging and effective, which spurs me to do more.
This week, as part of my self-study and for the Machine learning for trading course at OMSCS, I learnt
- Math behind gradient descent - sounds scary at first but its almost just applied pre-university math.
- What kinds of dataset works best for linear regressors vs decision trees. Pretty interesting and shines a light on how the models actually learn.