Michael Hammer

2023 November Newsletter

Hey!

MY QUICK UPDATES

I finally finished Andrew Ng’s Deep Learning Specialization. I’ll have a video review out soon!
I have also been having fun using ChatGPT to help me learn Spanish.

THOUGHTS, INSIGHTS, AND ADVICE

You don’t need to be an expert in each of the core math fields (calculus, linear algebra, statistics) in order to understand and use the basics of Machine Learning. When you’re just getting started you just need to know some of the main concepts. However, in order to develop true expertise in the field I believe we need to eventually understand the math more deeply.

RANDOM LEARNINGS, KNOWLEDGE, AND REMINDERS

The main way that derivatives (gradients) are used is to help Machine Learning Algorithms learn. Imagine you have a graph of the amount of error that your Machine Learning Model has given different parameters. Maybe it looks like a parabola (a U shape). The derivative is like the slope of a graph. In order for a Machine Learning Algorithm to learn good values for parameters, you want to move down the slope of the error graph. You want it to find parameter values that make the error closer to 0.
The main way that matrices are used is to do faster computations. For example, instead of writing code that loops over all the examples and parameters, you can use matrix multiplication to do all those calculations at once. This is often called “vectorizing” your code.

OTHER USEFUL/INTERESTING RESOURCES

Google’s Bard AI is similar to ChatGPT. I have been using it this past month and I really like it. I use it mostly to just ask random questions on my phone since it’s easy to log in with the Google App.

Michael Hammer