cicada2014

It really depends on the quality of your research experience and whom you are doing research with. The fact that there has not been a publication (even a workshop) after 15 months is bit of a red flag. To be honest I would not waste time applying to Stanford/MIT/Berkeley, unless you think you can get a high profile publication (first author) between now and Dec 15th. I would also cross Princeton/CalTech/Harvard off the list, since their AI/Robotic is small, and you aren't an "obvious" admit. Finally, it seems like you've just picked the top 20 schools + ivies from USNews without giving thought to relevant faculty doing research in your area. This is usually a bad idea. I suggest you narrow down your search, and include some schools outside of top 20. Even UMass Amhert/Harvard (which is the "lowest" ranked school in terms of USNews) is very very difficult to get into. Sorry if this is not what you want to hear, but this is coming from someone who saw firsthand how insanely competitive thing are in AI/ML last year (and it doesn't seem like this year will be any less competitive).

most MSCS are friendly to non-CS applicants (that's the point of the MS!) phd is a different story

Publications are hugely helpful, but they are neither necessary nor sufficient. - I had a first author publication at a top conference, but did not get into top 4 (I did get into others though). - I know of people who had no publications, but did get into top 4. They had superlative research experiences, though.

Columbia's MS in CS is not very well regarded (mostly seen as a way for foreign students to get Columbia name on their resume). Both Umass and UoE are extremely strong in ML. Even if you got into Columbia PhD in CS you it's still not clear that Columbia PhD is better. Against MS it's a no brainer.

Both are excellent schools in ML. Echoing what others have said, it will come down to whether you want a job in US or Europe after graduation. Faculty-wise UMass has Andrew McCallum while UoE has Ian Murray. Look at their publication records and see which seems more in line with your interests.

I've heard this year was particularly tough especially in areas like ML/AI which have gotten a lot of popular press. Many programs received record number of applications

Something to consider about Yale: http://www.businessinsider.com/yale-computer-science-petition-2015-3 I would go with Wisconsin.

Presumably if you are interested in a specific area, you have research experience in it, and thus you know at least one person working in that area (most likely your undergrad research advisor). He or she will be best positioned to give advice on which schools are good. Most rankings are (1) outdated, (2) do not reflect strengths across different areas within CS. This is especially the case within top 20 schools.

It seems like first wave of admissions are done for most schools. But as others have said, it's nowhere near April 15th.

I think the biggest factor is glowing recommendation letters from well-known professors in your area. Next important is research experience. (Of course, these two are intimately linked). In my experience most of my admissions came from schools where the LoR writer contacted professors and personally recommended me.

Same, just from POI.

Anyone hear back?

For big data, UW PhD is better than Princeton PhD. For UW PhD vs Princeton Master's, this shouldn't be a question. Go to UW.

I've found that general CS classes don't really matter--what really matters is classes in your domain of interest, and more importantly, research. I had virtually no "traditional" CS classes but had lots of math courses and research experience, and I got into a bunch of decent CS PhD programs.

You should check with the programs and see if you are on the waitlist. They are usually accomodating of such requests.