Message from C, C++ talks

July 2019

— Hell no! Programming machine learning models or deep learning neural networks in C++ must be hell itself. At best, you can look for some framework that does the job, but don't risk your sanity doing it all by hand. Btw, scikit-learn is easy to use and works quite good

Message permanent page

— Always use the right tools for the right job

— Oh, I don't really think so. In my opinion it's a good practice to write your own ML program from the basics and to understand how does it work internally

Message permanent page

— I deeply disagree with you. There's lots of maths and statistics behind those models that you'll need some months or even years to really understand them

Message permanent page


But as I mentioned earlier, it isn't the fastest way and in addition it isn't the right way if you just need programm and don't want to know a little bit more

— I'm doing maths and computer science at college, and trust me, it's not worthy spending that much time understanding them. In the end, you just end up trying many models and select the best one

Message permanent page

— Maybe ¯\_(ツ)_/¯
But I like to learn a lot of compicated topics :D

I think we're completely in the opposite camps about it

Message permanent page

— I like learning too, but in the case of machine learning, the real important part is the practice. Theory is okay to set a solid base, but rarely you'll need to understand it. Use your time wisely, it's not unlimited

Message permanent page

— You are right

— Is it possible to store lookup tables in the instruction part of the L1 cache?

— How can I express the positive infinity number in c++?

— Hi

— Guys anyone know about Data structure. ?!

— Yes?

— Std::numeric_limits<ull>::max();