Live data from Hacker News

Artificial Intelligence Graduate Certificate

scpd.stanford.edu

91–100 of 122 posts

Re: Artificial Intelligence Graduate Certificate

#91
post #27

Earlier quoted context omitted.

Outside of the big companies, ML/AI is often a solution without a problem. I would say just the opposite. There isn't a business too small to benefit from ML/AI. That is, assuming you acknowledge that there's more to ML/AI than "deep learning". Not everybody needs a deep neural network. Sometimes you just need linear regression or a random forest. If you come at it from that point of view, the knowledge you gain from…

> There isn't a business too small to benefit from ML/AI. That's not exactly what I'm saying. There are ML problems everywhere, opportunities to create efficiencies, new products, etc. out of existing data and processes. But nobody hires an AI/ML candidate to come in and tell them what to do. Nobody says "we want to dabble in ML somehow but don't have any pressing needs, can you start at $140K?" Those businesses need…

You might be be surprised.

Not all of the opportunities are reasonable (or make any kind of sense at all) but if you take a shitty hire at a place with interesting data you can have fun.

I can hardly count the number of organisations which now want to project the idea of being "innovative" and are excited to slap the coconut shells over their ears and try waving down planes with a stick.

There's still a surprisingly huge number of companies still hiring "blockchain consultants" even though they're rapidly heading into the "despair" phase of the hype cycle.

If you're a low cap stock or even a tired brick and mortar, two ML hires and a plausible investor story gives your graph a serious bump.

The difference between blockchain bull()$& and ML bull()$& that is the ML could actually have legs. Mostly it'll just tell you stuff that someone who understands market actually knew though.

Re: Artificial Intelligence Graduate Certificate

#92
post #85

Earlier quoted context omitted.

You might be interested to know that it's possible to enter a Master's program without completing a Bachelor's degree. This is not openly advertised, but I am aware that even universities with high signaling value like CMU and Oxford are open to this arrangement.

In general Masters in CS programs require either a BS in CS or coursework in areas like operating systems, data structures, algorithms, computability/complexity etc. These courses generally have prereqs such as 1-2 semesters worth of programming courses, discrete math etc. Someone without a CS background most likely would not be accepted to a standard MS program.

I am directly familiar with students accepted to CMU and Oxford for CS graduate programs (and other universities, but they aren't as noteworthy) without a Bachelor's degree. It's true that they had a CS background, but they demonstrated that with experience and one or two specific Coursera courses in areas that were lacking.

Re: Artificial Intelligence Graduate Certificate

#93
post #38

Earlier quoted context omitted.

According to another HN thread this is actually not true. Only having a certificate, without CS/AI undergrad, makes it almost impossible to get a job. You need Masters at least for most positions, some engineering positions you can get in with Undergrad in CS and taking the right AI courses.

Do people in the tech industry find any more value in physics/math undergrads than CS? CS is not as rigorous. I always thought that physics/math undergrad + ML certificate might be just as competitive as a masters in ML coming from a non-relevant BS

I'm going to have to disagree.

Physics/Math is useful but your particular field/speciality in CS is relevant as well. I would bet a bit more money on a CS grad to understand convexity, comp geometry and optimization than a physics grad, and math majors don't always have the code skills to develop things for production.

Point of argument here being frequently neither do CS Phds and at tech companies you pair Math/CS Phds with 'production' engineers.

Also certificates aren't really rigorous necessarily and also are quite easy to one-off versus relevant programs/theses/research experience.

production is an overloaded term depending where you go

Re: Artificial Intelligence Graduate Certificate

#94
post #85

Earlier quoted context omitted.

You might be interested to know that it's possible to enter a Master's program without completing a Bachelor's degree. This is not openly advertised, but I am aware that even universities with high signaling value like CMU and Oxford are open to this arrangement.

In general Masters in CS programs require either a BS in CS or coursework in areas like operating systems, data structures, algorithms, computability/complexity etc. These courses generally have prereqs such as 1-2 semesters worth of programming courses, discrete math etc. Someone without a CS background most likely would not be accepted to a standard MS program.

One of my colleagues dropped out of his undergraduate degree and went back to do his Masters in his 40s. Given his experience as an operating systems architect in the meantime it's not surprising he was accepted without a bachelor's degree.

Re: Artificial Intelligence Graduate Certificate

#95
post #90

Earlier quoted context omitted.

In general Masters in CS programs require either a BS in CS or coursework in areas like operating systems, data structures, algorithms, computability/complexity etc. These courses generally have prereqs such as 1-2 semesters worth of programming courses, discrete math etc. Someone without a CS background most likely would not be accepted to a standard MS program.

i'm sorry but this is not correct. CS Masters programs are more money-making then credential caring. Georgia Tech's MS as well as CMU's both accept non-CS traditional backgrounds. This is trivially verifiable on linkedin. That said they do require high ugrad GPAs frequently. Source: Knowing Georgia Tech Phd TA's and being a CMU TA for Master's classes

> Georgia Tech's MS as well as CMU's both accept non-CS traditional backgrounds.

There are exceptions but in general CS Masters programs look for sufficient coursework/background in Computer Science and Mathematics. An art major with 0 math/programming courses will find it difficult to be accepted to a Masters program in CS.

Re: Artificial Intelligence Graduate Certificate

#96
post #69
post #24

Earlier quoted context omitted.

I noticed that Andrew's course is still available. https://www.coursera.org/learn/machine-learning Is it still a good introduction to machine learning? It's several years old now. Are there better courses available?

It's a good course. The only(?) real criticism of it is the reliance on Matlab/Octave.

And even then, I can see why they use it - the syntax is pretty light and largely gets out of the way of the problem-space.

The only problem is that modern applied ML is dominated by Python...

Re: Artificial Intelligence Graduate Certificate

#97
post #90

Earlier quoted context omitted.

i'm sorry but this is not correct. CS Masters programs are more money-making then credential caring. Georgia Tech's MS as well as CMU's both accept non-CS traditional backgrounds. This is trivially verifiable on linkedin. That said they do require high ugrad GPAs frequently. Source: Knowing Georgia Tech Phd TA's and being a CMU TA for Master's classes

> Georgia Tech's MS as well as CMU's both accept non-CS traditional backgrounds. There are exceptions but in general CS Masters programs look for sufficient coursework/background in Computer Science and Mathematics. An art major with 0 math/programming courses will find it difficult to be accepted to a Masters program in CS.

Yes, of course. But that's moving the goalposts a bit. When I originally made my point, it was to another commenter here who is presumably a self-taught software engineer without an undergraduate degree, not an "art major."

For other commenters reading this thread down the line: my point here is that yes, while it's an exception, it isn't exactly stunningly difficult to get admitted to a good CS graduate program without an undergraduate degree. You don't need to be a prodigy who is so inarguably talented that you're just skipping the bachelor's. You'd be shocked what you can bypass by convincing a real human with decision making ability that you can do the work, instead of relying on every rote admissions page.

Admissions pages for graduate degrees are like job posts - almost all say they want an undergraduate degree at minimum, but when you open the kimono many are willing to silently drop that requirement without advertising it to others.

Re: Artificial Intelligence Graduate Certificate

#98
post #92

Earlier quoted context omitted.

In general Masters in CS programs require either a BS in CS or coursework in areas like operating systems, data structures, algorithms, computability/complexity etc. These courses generally have prereqs such as 1-2 semesters worth of programming courses, discrete math etc. Someone without a CS background most likely would not be accepted to a standard MS program.

I am directly familiar with students accepted to CMU and Oxford for CS graduate programs (and other universities, but they aren't as noteworthy) without a Bachelor's degree. It's true that they had a CS background , but they demonstrated that with experience and one or two specific Coursera courses in areas that were lacking.

My comment was aimed more for someone without a CS background entirely. If you have sufficient CS preparation, then of course you can be admitted to a MS program (assuming you satisfy whatever other departmental admission requirements). I'm sure there are exceptions out there/programs that accept non-traditional students. Even if you lack a CS background (assuming you already have an undergraduate degree), you can make it up in postbac programs and the like.

Re: Artificial Intelligence Graduate Certificate

#99

Earlier quoted context omitted.

In general Masters in CS programs require either a BS in CS or coursework in areas like operating systems, data structures, algorithms, computability/complexity etc. These courses generally have prereqs such as 1-2 semesters worth of programming courses, discrete math etc. Someone without a CS background most likely would not be accepted to a standard MS program.

One of my colleagues dropped out of his undergraduate degree and went back to do his Masters in his 40s. Given his experience as an operating systems architect in the meantime it's not surprising he was accepted without a bachelor's degree.

For posterity's sake, do you mind telling us which university it was? Or at least what "tier" university it was?

In my experience I've found (ironically), that it's generally the more reputable universities that are actually willing to do this, because "celebrity" (and I use that word very loosely) professors can talk a promising graduate student in over admissions department's veto at those institutions.

Re: Artificial Intelligence Graduate Certificate

#100
post #40
post #18

Earlier quoted context omitted.

The available classes align with graduate level one's for their MSCS. If nothing else it's markedly worse because it's 1/3 the classes :). Beyond that I'd expect rigor ... it's Stanford after all.

I am feeling that even a non-Stanford place would present more rigor than an online Coursera course. When I was in school, I benefited greatly from having 1-1 private conversation time with mathematicians, scientists, grad students, my friends, etc. With Coursera, that person is replaced by Google/Bing/Whatever. If I don't get time to talk to the scientists or grad students (at least), I don't see how Udacity/Courser…

When you pay for Udacity/Coursera/etc you at least get human interaction with paid instructors and other students.
Post reply on HN