How do I get started with machine learning? >inb4 Andrew

File: silicon-valley-ep3-table.jpg (45KB, 648x398px) Image search: [Google]

45KB, 648x398px

>>60604681
Your first project should be to determine the differences between a hotdog and a penis. Spend the next few months of your life looking at hotdogs and penises on the internet while reading some machine learning books.

>>60604681
>Hi guys lol I want to create skynet what could possibly go wrong.

>>60604707
Is this extra credit

>>60604681
>How do I get started with machine learning?
Why? There's a thousands of other retards like you already "getting into it". You're never gonna accomplish anything anyway. Just go back to watching anime.

>>60604782
>>Hi guys lol I want to create skynet what could possibly go wrong.
Basically every single country that matters is racing for exascale for AI research desu.

>>60604681
You take stats basically

Here's some resources and lectures
https://functionalcs.github.io/curriculum/#org5e79eb7

Remember 'computer vision, machine learning, deep learning' ect are all just stats.

>>60604812
They're all busy making driverless cars, nobody's doing algos for automatically sorting animus.

>>60604851
Also, I forgot to mention they don't call it ML anymore, it's now "AI engineer" as the preferred buzzword.

>>60604851
This. I call it computational statistics.

Geohot is looking for stat 'engineers' http://comma.ai/positions.html

Prob pays shit but Willy Wonka Elon Musk is undoubtedly going to buy that company for a few billion might be worth it if the job offers options

course.fast.ai

https://work.caltech.edu/telecourse.html

This is a real ML course focused on theory. People who say bullshit like "ML is just statistics" or "ML is just linear algebra" know just the bare minimum to be ignorant about their own ignorance, don't listen to these people.
This is not an easy course (aimed at people who are at the end of their undergrad or beginning their masters) but the teacher is just so excellent you can get away with a lot less knowledge.
You won't be building pretty-but-useless shit like in Andrew Ng's course but you WILL have a solid understanding of the basic theoretical principles behind machine learning and should be able to make your own way regarding what and how to learn.

Good luck OP.

>>60605348
Uh, machine learning is just stats bro. The standard text in ML, "The Elements of Statistical Learning" is authored by statistics Professors.

>>60605541
FFS... Machine learning covers a lot of areas, and while it is mostly stats, linear algebra and calculus it definitively isn't "JUST stats" or "JUST linear algebra".
Statistical Learning is just a subset of machine learning. It covers machine learning techniques rooted in a more traditional statistics approach, but it is very far from being all of machine learning.
If you still think that tell me, how much statistics is there on traditional perceptron based neural networks? Certainly a lot of calculus and linear algebra but very little in the way of statistics.
Trying to dismiss a whole field as "meh, it is just a new name for X" when you clearly are ignorant about that field makes you look stupid. Please don't do that in public, you won't be taken seriously.

Welp, guess i'll be going Intel from now on.

>>60604707
why did they take a dump on those people working in the data center offices? Is it really so horrible working in a data center?

>>60605586
It's literally stats.

https://statistics.wharton.upenn.edu/programs/phd/

>>60605846
Stats department studies ML, so the logical deduction you make is that all of ML is contained within stats?

>>60604991
I am willing to bet against this. Musk's in house automated driving research team is a million times better than geohot's hot garbage.

>>60605926
Not about the technical skills themselves.
Buying them prevents competition, makes for a good PR move, most likely affects stocks, etc.

>>60605876
It's just stats with different marketing.

Here is the top researching in ML, as we shall appeal to authority: https://normaldeviate.wordpress.com/2012/06/12/statistics-versus-machine-learning-5-2/

>What is the difference between these two fields?

>The short answer is: None. They are both concerned with the same question: how do we learn from data?

>>60605967
This, plus Elon Wonka has that sweet government funding in the billions to make aquisitions.

>>60604681
But OP
You can just watch Ng on double speed

>>60606046
Beg to differ with him. The objectives of both fields may be similar (statistics is not limited to just learning from data!) but that does not mean they are one and the same.
The tools and methods used are VERY different and that does make for a different field, even if in recent times they are growing closer together.
One simple example are neural networks, they aren't really in stats. Sure, statistical methods have been applied to help understand and develop them, but the way the came to be, how they work and so on is very different from what statistics usually study.
If you go deeper you can see more differences such as mixing computational theory and machine learning which has no real analogue in stats and so on.
So, IMO, they are very distinct fields. Sharing some common goals don't make both disciplines one and the same.