Am I being too optimistic about CS PhD application?

[wagyubeef]

Hi guys. I'm currently getting a research mscs in a top10 university in US (top10 cs undergrad in the states as well, both rankings are major rankings), and about to apply for CS PhD this fall. Here's some info:

- GRE: 161 verbal, 170 quant, 4.0 writing
- intern: 3 summer interns in the bay area
- research & publication: 1 first author at ISIT, 1 second author at ACL, 1 second author at EMNLP, 1 third author at EMNLP.
- GPA: undergrad 3.84, grad~3.9
- recommendation letters: my advisors are pretty good, although they may not write super strong recommendations

I would like to do research in learning theory, also tcs or mixed integer programming. Here's my list:
- reach: UCB, Stanford, MIT, CMU, Princeton, TTIC
- match: UT-Austin, UWashington, UPenn, GeorgiaTech (ACO program), UWaterloo (ACO), UCSD? Cornell?

My concern is that, although I have 4 papers, only 1 is about learning theory, the other 3 are about NLP, which I'm not sure whether would help with applications for learning theory direction. ML is getting very very popular nowadays and I know ML applicants who have a lot of papers get rejected by all the universities they apply... On the other hand, is learning theory a little easier to get in compared to ML? Since the former is theoretical. Also does anyone know how hard it gets to change research topics? I have no research experiences except for some courses (and reading on my own) in TCS or MIP, but they really appeal to me.

So am I being too optimistic about these choices? Thanks a lot for any suggestions!

[wagyubeef]

Anyone?

[iwearflowers]

I can’t speak to your field, but generally my sense is that committees understand that your resume may not reflect your interests perfectly. It’s more and seeing that you have broadly relevant knowledge, skills, and experience. Your SOP should explain your interests and why your experience has led you there.