Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Technically incorrect! The numerical chain rule uses the method of finite difference, which will accumulate errors as you move through the computation.

See the 'differences from other methods' section: https://en.m.wikipedia.org/wiki/Automatic_differentiation

The point, as the neighborhood comment says, is that the implementation really matters, and is worthy of study. It's fine to say that autodiff is a family of methods for implementing the chain rule, but incorrect to say that it's 'just' the numerical chain rule.



by "numerical" i meant "operating on numbers as opposed to symbolic values", not "using the method of finite differences", apologies for the confusion




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

Search: