What math does one need to know to fully understand all the cutting edge concepts of machine learning, to be able do do research in machine learning?
I'm talking about fields like kernel methods and neural networks with the recent boom on deep learning.
_action input
_mutate x100 variant
_select best output
//repeat
congrats. now you have a machine that learns through trial and error and goes with the best option until it figures out how to perform a task.
>>7658030
I thought computer science was for bad math students that still wanted to make bank.
>>7658030
Probability theory, linear algebra, topology, and some amount of of differential equations. Really though you're better off studying machine learning itself, most of the content is baked into the methods of the field itself. At least in my opinion.
> Probability theory, linear algebra, topology, and some amount of of differential equations.
You forgot statistics, unless you don't consider it math, which I would tend to agree with.
CS requires rigorous proofs
Engineers and most pure sciences just plug shit into equations like retards
just more evidence that most people on 4chan have no idea about the field they are talking about.
you should at a minimum have a very strong grasp on:
calculus, linear algebra, statistical analysis, probability theory, and differential equations. along with understanding basic data structures/algorithms and understanding runtime analysis
>>7658123
>unless you don't consider it math, which I would tend to agree with
>I only know the t test and the chi squared test
Edgy
>>7658141
statistics blows and you know it
>>7658128
Lol you mean proving the sum of two even integers is even in discrete math?
that shit doesn't count you know
>>7658244
this.
>>7658037
Define mutate
Define "best output"
>>7658431
define "define"
>>7658030
Most machine learning books require a really good understanding of linear algebra (basics, proofs, and even the numerical aspect) along with probability and stats.
Not much else really.
>>7658136
when do differential equations pop up? Never have I seen a differential equation in ML books
>>7658030
You don't need any of this if you are going for automatic reasoning based machine learning.
>>7658717
I didn't really see any as well in my course.
The only thing I can think of is when you add a "momentum" term to help find a better maximum to your optimisation equation. If you have no idea of how different derivative terms behave this might be a bit difficult to understand.
I'd say the statistics,probability theory, Linear algebra, abit of calculus, programming, basic comp science.
>>7658244
It only blows if you don't understand it.
The confidence interval in my above statement is true with a 99.9999999999999999% degree of accuracy. +\- .000000000000000001 margin of error.
