LACs for a student focused on AI/ML (double majoring in CS+Statistics)...

College Search & Selection
1

I am trying to get my son (finishing his junior year in large public high school) interested in LACs. Background: He is really only interested in CS (particularly ML/Deep Learning) and Statistics/Applied Math. He has taken all the STEM AP classes his high school will allow him to take, and has taken/(or will complete by the time he graduates from high school), all the lower division undergrad math classes at the local community college. In his spare time, he has been working with researchers at 2 different universities on projects at the intersection of AI/ML and medicine (they are getting research papers submitted to conferences/journals). He is pretty strongly in the camp that he only wants to apply to CS research universities. I am trying to get him to look at LACs (pointing out the focus on teaching, close contact with professors etc, that would be beneficial both as an undergraduate experience and when applying to graduate school).

Questions about CS programs at LACs (limiting to LACs which also offer a major in Statistics/Data Science) :

  1. Based on what I read in the forum, CS seems impacted even top LACs. They restrict the number of electives CS majors can take, freshmen are locked out of the gateway CS classes etc. I hate to name colleges here, but I’ve read about problems with CS impaction even at storied LACs like Swarthmore, Haverford and Williams. Even Harvey Mudd seems to be feeling the strain, because there are many students interested in taking CS classes and they cannot hire faculty. Are there LACs where CS is not impacted ? He will be extremely unhappy if I steer him to an LAC where he locked out of the gateway CS courses (which are pre-requisites for other CS classes).
  2. Which LACs offer the widest selection of CS electives (other than Harvey Mudd) ? Particularly in AI/ML ?
  3. The CS programs in most research universities allow majors to select a specialization track, and they take electives in that specialization in their junior and senior years. I wonder if there are LACs where one might get a similar experience ? It is likely too much to expect an LAC to offer a specialization track within CS.

Maybe the right answer is to apply just to the U of CA campuses, and the large out of state public universities (like most of his peers at his school who apply for CS) ?

2

I’d be more worried that the LACs can’t offer enough math and CS classes to challenge him if he’s already topped out of the math being offered to him in hs and community colleges. We looked at some of the smaller schools for my daughter who didn’t know exactly what she wanted to major in but something STEM and one school had 5 math professors. FIVE.

I think he’d do great at the UCs, but if you want smaller, look at the smaller STEM schools like Harvey Mudd, Colorado School of Mines, Rose-hillman, CMU or an LAC with a consortium like Amherst where he could take courses at UMass.

3

As the parent, you have done the financial planning to know what cost would be affordable, right?

Be careful not to overreach with the UCs. Some applicants apparently get deceived by exaggerated weighting of HS GPA which may lead to unwarranted optimism when compared with HS GPA listed on UC web sites, which are weighted (and capped) in a way that gives lower GPA than typical weighting methods at high schools.

http://talk.qa.collegeconfidential.com/university-california-general/2127392-faq-uc-historical-frosh-admit-rates-by-hs-gpa-2018.html can give you an idea of how competitive each campus is (with links to how to calculate HS GPA for UC purposes). But be aware that most campuses will be more competitive for applicants to the CS major than for the overall campus.

4

If he wants statistics and machine learning, I might not pick Colorado School of Mines, although its hiring new CS faculty all the time. Its a public school that focuses on all geology related fields, as well as mechanical engineering, and chemical and petroleum engineering, and they lost one top math/statistics faculty to CU Boulder recently. Its not an LAC, though and it is a public school, but students are almost 100% engineers, with a few that might study geochemistry or applied physics. The CS and applied math are growing, so that program is getting stronger and stronger.

Most top Colorado CS students prefer to study at CU Boulder, which is somewhat stronger in math and CS
compared to Mines, with more research programs, and bigger more active campus , with more distractions than the smaller Mines campus in Golden.

Given that you can get a better price at U of California campuses, assuming he can get in, then I don’t see
why one would choose a more expensive Colorado public school for CS, look at UC Davis, UCLA, SD and Santa Barbara are all ranked much higher than Colorado public options. Cal Poly SLO if he wants the large number of labs and hands on learning style. . I don’t like UC Santa Cruz as well as the other four,
but many out there love it. Santa Cruz has a newer engineering program and no mechanical engineering degree,
with if he were to change his mind, I think Santa Cruz is missing key classes for those that might want to study computer aided design. Of course UC Santa Cruz offers more engineering than most of the east coast LACs like Haverford or Williams though.

What about mid sized private research oriented schools ?

Look at Case Western Reserve U, its got a strong liberal arts focus,(see Western Reserve College ) and a very strong computer science and math program, with some research, but undergraduate focused. Phi Beta Kappa is valued, and every Case student reads a freshman book the summer before freshman year. Nice balance.

While its also not an LAC, Carnegie Mellon offers strong liberal arts classes, and a statistics/ML bachelors program in Dietrich College of Humanities and Social Sciences, that is strong, but not the same degree as their high powered School of Computer science. Dietrich has a lower admissions standard than SCS. Not saying your son could not get into SCS but maybe the Dietrich program is more LAC like.

http://coursecatalog.web.cmu.edu/dietrichcollegeofhumanitiesandsocialsciences/departmentofstatistics/#statsmltext

RPI in Troy New York is very undergraduate focused, but still offers some PhDs and masters degrees, very strong connections to IBM research, IBM supercomputing, IBM investments.

Also maybe Notre Dame University, Tufts University, Northeastern U. They are not all created equal but good CS and stats classes.

5

I wouldn’t write off LACs, especially ones that have strong STEM programs like Williams (the math department is especially renowned). As a prefrosh without any CS experience, I was able to talk with a prof from the CS department regarding taking intro classes. He told me that as long as they sign up during freshman year, students will be accommodated into all intro classes; they give freshmen priority. After getting those classes out of the way, you won’t have any issues signing up for higher level ones.

The real issue with the CS major at Williams mainly affects people who sign up after their first year.

6

I would also agree that without PhD level opportunities on campus, your S may run out of courses. I second looking at the smaller research universities.

7

strongly recommend New College of Florida, they recently started a master’s program for data science and their focus in their undergrad computer science is strongly in machine learning. I actually know people in the program, they are very, very intelligent students and I genuinely think it would be a good match. SO many opportunities for hands-on research at this LAC.