Ask HN: Is it important to know backpropogation when studying ML/ Deep Learning?

3 pointsarunsivadasan2 comments

I could understand how in an earlier age, knowing backprop was quite important. But is there any real world use for this now that most of these aspects have been hidden from a practitioners view via tools and frameworks like Tensorflow, Fast.ai etc

View on HN

As any subject to learn, you don't have to if you don't want to.

More seriously, as in programmation, many new languages manage memory with a garbage collector or other mechanisms, but if you want to be a good developer, you need to know how all this is done underneath.

To drive a car, you don't need to know how an engine works, but to create one...

And for the particular subject, you have to understand the principles under your tools to know how to change parameters and understand why the tools can be biased with bad parameters (or not enough data).

I think it would be more important to have a broader understanding of neuroscience

Reply on news.ycombinator.com