Hacker News new | past | comments | ask | show | jobs | submit login

The main thing to understand is the basics. For example element wise multiplication vs full blown matrix operations. Think something like this: http://en.wikipedia.org/wiki/Matrix_multiplication

Matlab/Octave is a great way to practice this due to the native data types. If python is your thing numpy's arrays are also pretty easy to digest.

Subtle little tricks like this: https://www.youtube.com/watch?v=evF-3ykjRU0

And understanding the dynamics of scalar operations vs matrix - vector operations.

The machine learning class has some good fundamentals if you need a refresher on how something works.

There will be more complex things like some optimization algorithms have different uses for eigen values: http://see.stanford.edu/materials/lsocoee364b/11-conj_grad_s...

See: http://en.wikipedia.org/wiki/Eigenvalues_and_eigenvectors

One other thing might be understanding different ways you can manipulate data. In this case, numerical representation here is an example per row when I toss in one matrix for training. This is applicable to many machine learning problems.




Very much appreciated. That's really helpful and feels very doable. :)




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: