Hello!
I’ve completed Machine Learning Specialization. I understand the fundamentals of ML such as Supervised and Unsupervised Learning. But to actually implement all these concepts and if needed in order to understand the maths as well, what should be my next step please guide?
I am currently a Software Engineer with 8 years of experience. So Python isn’t a concerning end for me. I am more concerned about implementations of ML and Maths.
*As to the maths question, I’ve been doing this somewhere else, but even for me I think it is useful to see ‘what are the problems we are dealing with’ first, then you can go back and attack the maths.
The mathematics isn’t very important if your goal is to use existing tools such as tensorflow. They provide all of the math under-the-hood. You design the network, it handles the calculations.
If you want to invent totally new methods, or if you want to implement models for platforms that don’t support advanced tools, then yes you’ll need a strong math background.
The tricky math in machine learning is in training the model. It requires some calculus to get the equations for using the gradients to find the weights that give the minimum cost during training.
There are tools that already implement the required training methods. If your platform supports those tools, it’s fairly simple to implement.
ML also involves some simple statistics, but that doesn’t require special math skills.
@TMosh@icybergenome for better or worse, I’ve gotten involved in this and I suggest you check out Andrej Karpathy’s series of videos, at least up to Micrograd.
This should at least give you a practical understanding how in the ‘real world’ (i.e. discrete computerized systems) this is handled.