What
Machine Learning models is a fancy way to say
Mathematical Formulas that solve one or more problems, and were not created by humans.
Why
In most cases we use Machine Learning Models to automate tasks, but they can also be used to structure messy data. Here are a few examples:
1. To Automate Tasks
Say you have a task that you keep on doing on and on that muscle memory is enough and you don’t even need to take decisions anymore. To get back some of that time, you can build a ML model to do the task for you.
2. To Structure Messy information
Similar to how in this blog I’m trying to structure all information I gathered from all the places I learned about Machine Learning and Game Development, a ML model may be created from data that isn’t structured the same (reddit, YouTube videos, word of mouth). ChatGPT is the best example of how you can build a ML model that has found a structure of a lot of data on the internet, and now can answer questions from pretty much any domain.
How
Let me clarify that, humans don’t design the actual Machine Learning Models. They build Machine Learning Algorithms that learn from examples and produce a Machine Learning Model that describes the solution.
Here's an Example
Imagine you’re the Machine Learning Algorithm and know nothing about baking cakes. Then I give you a bunch of ingredients and the cake you must bake.
Your job is to come up with the recipe of how to bake the cake, that then can be followed by anyone given similar ingredients (maybe different ingredient brands).
The cake recipe is the Machine Learning Model.