Is machine learning a meme or is the foundation for a new frontier?

livemint.com/Sundayapp/zDSjhU5IzcuI7ypo6W4WtL/Why-data-science-is-simply-the-new-astrology.html

I don't agree with his arguments but I kek at the title

Thanks, Google, I totally trust you now.

;)

Machine learning is modern statistics. That's literally all it is. It's extremely useful but it's not magic.

Right now it's a meme. What researchers call 'machine learning' right now is really just advanced statistics. We are nowhere close to creating a general purpose learning algorithm on the level of humans.

For the record, I am only responding to this post because those girls are sexy.
That's how we deal with malware.
Ever used a gameshark?

When you complain about hype in excess, you are generating hype by contrast. Thus, your complaint ironically makes you a hype-man.

The extension of your criticism is that people put too much faith in ML as a field of its own, and not enough trust in fundamental Mathematics, Computer Science, and Electrical Engineering. There is still a good amount of money to be made in ML, if you treat it as approbation of applied math- then it creates economic efficiency where it replaces human error.

Your ordered list need to be recast as a venn diagram. Private research and small companies have the best market value, whereas undergraduate degrees and big companies have the most horribly inflated value. Graduate students are all over the spectrum, and the best thing you can do is to classify them by age. A 28 year old with a PhD got lucky and was favored by university nepotism. A more senior researcher is always going to be more valuable, because a majority of implementation is always going to be based on extant infrastructure and systems.

This is NOT the end game of ML but Veeky Forums fags will tell you that people think it is in order to discredit ML

Jumping headfirst into a Wikipedia article does not compensate for a lack of training in graduate level mathematics. Try reading about the Borel Measure before you attempt to interpolate what is meant by the word 'learning' in this specific context.

No, we already had genetic algorithms with evolving control structures. It seemed like that research slowed down after I got plagiarized by a private school guy who misunderstood my lesson about the exponential family of distributions as an equivocation between NNs and GAs. Short summary: Georgetown University completely fucks up the University of Maryland, because it's a bunch of rich southerners who only come north of the Potomac to rob people.

Why don't you go look into proof assistants.

>This is the first stepping stone to telling computers what we want them to do

It's not the first, because there was a long time without any visual interface.

>A computer doesn't truly know what an object is.
>A computer doesn't truly know what an object is.

We're reliant on semiotics to state what is truly known. Semioticians rely on mathematics for objectivity. If you want to push philosophy up the heirarchy, then you have to restrict it to the design of language and grammar.

>What researchers call 'machine learning' right now is really just advanced statistics

No, "advanced statistics" is the name of the course that they make people take at the private school diploma mill. "Digital Analysis" would be harder to hijack for a buzzword.

It isn't the endgame, because we still have to improve that machine. Even if the machine can improve itself, the notion is stored in a bounded value and subject to a resource constraint.

no one can and it will never happen
it's all bullshit, no matter what any jackass says