Udacity was born due to the popularity of this course.
Can't thank Sebastian and Stanford enough for this free course.
edit: People below (and above) are more knowledgeable than I am.
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?
* edit: I just learned the courses were simultaneous.
Udacity launched June 2011, Coursera April 2012, according to Wikipedia.
I'm not sure which MOOC company started first (in 2012), though - I do think it may have been Udacity. I know that when it did start, it wasn't able to offer the AI Class because of {reasons} (I think maybe licensing of course content or something) - and so instead it offered the CS373 course (at the time titled "How to Build Your Own Self Driving Vehicle" or something like that). By that time (Spring 2012), I had moved on past my earlier problems, and jumped at that course - I took it and completed it as well.
Coursera, meanwhile, was able to offer the original ML Class as a premier course (maybe Ng had different rights to the content, or maybe there was issues on the AI Class that had to do with the dual instructor partnership of the class - I'm not sure what really happened there).
After about a year (IIRC?), Udacity was able to finally offer the original AI Class as part of their courses (I think now renamed "Introduction to Artificial Intelligence").
Today, I'm taking the Udacity Nanodegree course, as I've noted before here.
I think this offering from Stanford is interesting, but it doesn't seem like it is currently "available" to enroll, because the two required courses seem closed or something? Maybe they're taking enrollment for a future starting date. That said, while the tuition isn't outside of something I could do, I currently think the best use of my money and time - after I complete the Udacity thing - would be to pursue getting a BA on CompSci or something of that nature, then pursuing other paths.
Should've done it a long time ago when I was younger - but I was dumb.
The question is only whether the HR person's neural network triggers more strongly on "master's degree" or "stanford".
None of this has anything to do with what you learned or know. Like most tertiary education.
So for jobs, ML/AI is pretty much an even field unless there are niche research jobs which only requires Phd gradudates
[1]https://github.com/antirez/neural-redis
[2]https://twitter.com/Mikettownsend/status/780453119238955008
http://techcitynews.com/2014/08/07/big-data-is-a-lot-like-te...
And I think what I perceive as your doubts as to landing a job with one of these are probably founded. This is a way for the university to make a bunch of cash. As long as that happens, whether students get a related job or not is not important to the university except to help with marketing to prospects.
https://www.edx.org/micromasters/columbiax-artificial-intell...
(EdX employee)
So far SDC is the best fun I had in a while, getting a car drive all by itself on a circuit feels absolutely cool! ;-) Have fun as well, I hope they have more cool stuff prepared for us!
By MIT self driving course are you referring to http://selfdrivingcars.mit.edu/ ?
I'm not sure their brand name justifies that price (not sure about the content). The competition is probably the AI nanodegree from Udacity which costs 800$/term with the chance to earn some of that back. If the employer I want the certificate for knows what online certificates are, chances are they are familiar with Udacity (possibly more so than with Stanford in that market).
They want to charge 20k, but not let anyone have a chance of further advancing to complete a real degree, no matter how excellent their performance in this program.
The reason they do this is solely to protect their brand and exclusivity. They already offer online degrees but the acceptance rate is just as limited as the on campus program.
Yes the learning is important, but so is the credential and a certificate doesn't even come close to a degree in the job market.
Stanford should pick one:
1) Charge Stanford prices, scale up online, and let any student who can do the work pay tuition and earn a degree.
2) Charge lower prices for certificates and continue to artificially ration real degrees.
And so on. It was overall a pretty negative experience for me. I can't recommend it.
So? Do they say that it does? I'm already a software engineer and can tell you that having an AI certificate from an accredited University is a great stepping stone to transitioning into this line of work even if it does not make me an expert.
Okay, graduate degrees fulfill that purpose too, but with life events now and over the horizon, I just don't have the bandwidth to commit to getting my Masters or a Ph.D. right now.
I really have no problem with this.
Depending on hiring filters, that might be true. If an application ever gets to an engineer though I'd bet on the last one or two winning.
I think a sensible organisation should skip the usual hiring filters in new fields, because [edit](my experience is in security) you can scoop up really good people who happen to have "unconventional" backgrounds if you have competent people evaluating them.
Regular HR people tend to do a really bad job with career switchers and the self taught since they mostly work off "signalling value". But in newer or very fast moving fields, the oddballs can be majority of decent applicants.
My limited experience is that very technical positions are not actually overwhelmed with applicants, and that it's not hard to evaluate if people have the right stuff because in these areas it's not difficult to devise quite objective challenges without resorting to shibboleths (guess what the interviewer is thinking or, do you come from the same technical culture as me).
Arguably ML positions should be the easiest to algorithmically hire for (at least for industry, not hard core research). Just put an automated judge with a fairly low bar on a business relevant objective function between your careers and your "submit job application" page :p
Personally I find this a little bit funny, because ML driven competence evaluation in "hard" (reasonably concrete objectives) fields should eventually render credentialism and signalling obsolete. But here we are, the $20k "certificate".
All these things said, a structured course of study is super useful for the undisciplined (certainly including me) and dropping "new car money" on something has a way of focusing the mind :)
That indeed does not exist, but if you do well on these, you are probably pretty eligible for their part-time MS program.
Moreover even if it increases your chances a bit, it's a massive investment with no guarantee. Why not simply let the students who demonstrate excellence in earning the certificate a chance to spend another 50k for a degree?
The reason is to maintain artificial scarcity. Stanford has done a lot of great things and all the grads I've worked with have been top notch. However in some ways, Stanford is to education as De Beers is to diamonds.
The math is mostly applied and not proofs. There's a fair amount of pseudocode and algorithms but they are explained well and I think it's not hard to follow (our students of different backgrounds usually didn't have problems). I did get a bit tired of the running example of the map of Romania (essentially used for all the search related things). The diagrams for algorithms are very helpful.
Whether that provides more value or employment opportunity I cannot say, but it is encouraging to see more to universities offering an alternative to traditional degree programs.
I think there's a path for non-MS, non-PhD backgrounds, but probably not now. Outside of the big companies, ML/AI is often a solution without a problem. So until they learn practical application I think supply will outnumber demand and most of the jobs will go to PhD AI/statistics backgrounds.
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 taking Andrew Ng's machine learning course is enough to create value for companies / organizations.
A lot of the discussions on this topic here on HN seem to be based on an assumption that you have to be doing cutting edge original research to be useful. I think that's very far from the case.
Now the problem is, does the mom and pop bakery on the corner know that they could benefit from ML? And perhaps the more salient question is "are they looking for somebody to do ML for them"? The answer is probably "no" in both cases, so you might have to do some work to sell to that market. But the value you can create is real. Help them optimize production so they throw away less bread on slow days, etc. and you're talking direct business impact.
I suspect that there will be, for a long time, a wide continuum in terms of how ML/AI skills can help create value. Which means there will be a lot of ways to leverage this field from a career standpoint.
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 to recognize the opportunities first before the market will expand.
Based on my experience, which I won't re-iterate here - the various MOOCs I've taken (and currently the Udacity Nanodegree) would not be anywhere close to a masters in the subject (unless I am severely overestimating a masters - but I don't think so).
TBH - they would probably equate closer to an Associates, at best.
This offering from Stanford? Not sure - but I still don't think it would be the equivalent. I'm not saying it wouldn't be worthwhile, but I think if your goal is a deep level of knowledge and understanding of the subject, then a quality masters program for CompSci or similar would be the better path.
I'm now working in the ML/AI research division of big-4, all I needed (and all hiring teams needed to see on my resume) were a couple of these classes (with good internal performance ratings).
Having said that, my employer paid for these classes. No way I would pay this price out of pocket. There are probably much cheaper ways to get the same knowledge.
You actually study, partner with, and test along-side Stanford MS/PhD students. Not to mention the accountability factor, since you paid $5k for the class you better take it seriously.
I have interest in AI, but I'm not sure how companies would see this online degree (even through it is from Stanford).
Does a "solid foundation in AI" actually exist?
I'm asking because it seems that nobody really knows why many algorithms actually work, or even how they should be adjusted to cover new applications. To me it sounds more like "educated guessing".
with a MOOC, there doesnt really exist the same social pressure to continue or excel.
Its been interesting to watch how various MOOCs have tried to recreate these pressures (due dates, courses separated into weeks, peer review assignments, teacher 'office hours' etc.) and while i think they've gotten a lot better than at the beginning, im not sure they will be able to fully replicate the pressures of having real people in the classroom with you, who will notice if you are gone for a week
https://www.edx.org/micromasters/columbiax-artificial-intell...
-------------
Who is this MicroMasters Program intended for?
The MicroMasters Program in Artificial Intelligence
is intended for those who have a Bachelor’s degree in
Computer Science or Mathematics and have a basic
understanding of statistics, college level algebra,
calculus and comfort with programming languages.
-------------
-------------
Complete, pass and earn a Verified Certificate in all
four courses to receive your MicroMasters Credential.
Learners who successfully earn the MicroMasters
Credential are eligible to apply to the Master of
Computer Science program at Columbia University.
-------------
https://www.udacity.com/course/artificial-intelligence-nanod...
Also - is this closer to a Master's level program or part of an undergraduate curriculum?
I'm enrolled in Stanford's CS Master's program right now through the Honors Cooperative Program (which lets you get a Master's online while working in industry), and I'm currently planning on doing a dual specialization in Systems and AI. For the AI specialization I've already taken CS 221 and 229, and I'll have to take three more AI classes drawn from a list pretty similar to the Elective Courses list in the OP.
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.