Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Oxford Undergrad CS Admission Interview Problems

comlab.ox.ac.uk

Post reply on HN
Print view View original

21–30 of 52 posts

Re: Oxford Undergrad CS Admission Interview Problems

#21

Post by sachinag »

andrew1 wrote: post #13
sachinag wrote: post #5

Could some kind UKer(s) please explain to this American exactly how Oxford, Cambridge, and the UK university system works? I'm familiar with what A levels are, and that you supposedly apply to a particular college at each of Oxford and Cambridge, but then I hear about university-wide "departments", and I put that in my American-style mental model. Googling hasn't helped me, but perhaps I'm being unimaginative in sear…

Only some UK universities have a collegiate system (Oxford, Cambridge, Durham, St. Andrews(?)). Most do not (I believe) and I imagine are roughly similar to US universities. At collegiate universities there is a split between what a college is responsible for, and what the university is responsible for. Lectures are university-wide, exams also are (you get your degree from the university not the college), departments…

OK, this really helps; thanks! Let's say that you really wanted to study CS at Oxford; if you pick the wrong college to apply to, are you shut out from the university entirely? Or can you apply to multiple colleges at the same university at once?

Re: Oxford Undergrad CS Admission Interview Problems

#23

Post by regularfry »

smcl wrote: post #8

I suspect the Oxford\Cambridge undergrad interviews have less to do with assessing technical competency and more to do with providing admissions with a reason to discriminate against state-educated kids. The amount of Oxbridge intake come from private schools is disproportionately high to be due to exam results alone, and is better explained by an ingrained negative view of those with a state education or old boys ne…

That's likely to depend hugely on the subject. I'm privately educated in relevant subjects, and went on to do an engineering degree at a well-regarded university (not Oxbridge, but ranked with them), but there's just no way I had the technical chops at the time to get past these questions. I don't think I'd even read GEB yet.

With a technical subject, it's relatively straightforward to put a meritocratic admissions system in place. I don't think you can make that argument for humanities subjects.

Re: Oxford Undergrad CS Admission Interview Problems

#24

Post by apl »

smcl wrote: post #8

I suspect the Oxford\Cambridge undergrad interviews have less to do with assessing technical competency and more to do with providing admissions with a reason to discriminate against state-educated kids. The amount of Oxbridge intake come from private schools is disproportionately high to be due to exam results alone, and is better explained by an ingrained negative view of those with a state education or old boys ne…

All Harvard kids are rich rowers, all Oxbridge students are posh sons and daughters of politicians and aristocrats, and all these state-educated kids were rejected simply because they didn't wear the right clothes, amirite?

Bla.

I do around 50-60 admissions interviews a year, and we don't even know to which schools the applicants went when we question them. What you've seen in action is a few anecdotes and whiny parents who blame external factors for their child's rejection letter.

NB, I'm not saying there isn't a problem. There's still a lot of room for improvement; both universities are doing plenty of outreach work and are fully aware of the situation. But your portrayal of Oxbridge is ridiculous.

Re: Oxford Undergrad CS Admission Interview Problems

#25

Post by jacobolus »

viraptor wrote: post #9

I really like those questions - they seem very tricky until you think of some way to do it in a smaller subset or try to figure out the solution by doing the actual actions in your head. I think they're very different from the typical hiring interview logic puzzles you'd see in the companies, but I can't really express the difference... Then again, I can't really figure out (an optimal) solution to the second one...…

Question #2 also doesn’t clearly state whether we learn the value of each guess before making the next guess.

As for your idea, 1/3^5 accuracy is not correct. You need to think a bit more carefully about what you can actually learn from testing a pair of points.

Re: Oxford Undergrad CS Admission Interview Problems

#26

Post by jacobolus »

JoachimSchipper wrote: post #16
viraptor wrote: post #9

I really like those questions - they seem very tricky until you think of some way to do it in a smaller subset or try to figure out the solution by doing the actual actions in your head. I think they're very different from the typical hiring interview logic puzzles you'd see in the companies, but I can't really express the difference... Then again, I can't really figure out (an optimal) solution to the second one...…

I don't really see what you mean ("3 segments" suggests asking for f(1/3) and f(2/3) to me, but that's not enough data to know anything). I'll type what I thought, which may be the same thing. If a = f(c) then a You could recurse on the above, e.g. by measuring 1/4 + 1/8 and 1/4 + 3/8 if 1/4 Another interesting trick is measuring 1/2 + epsilon, which (almost) bisects the range with only two measurements. [EDIT^2: rem…

Measuring 1/3, 1/2, and 2/3 does not tell you which 1/3 of the range you should look. There are 3 cases to consider:

1. f(1/3) is the biggest. In this case you learn that m is between 0 and 1/2.

2. f(1/2) is the biggest. In this case you learn that m is between 1/3 and 2/3.

3. f(2/3) is biggest. You learn that m is between 1/2 and 1.

* * *

Edit: as for measuring 1/2 + epsilon, it’s no worse for “adversarial input”. You just need to remember that your first pair of measurements tell you that either 0 ≤ m ≤ .5+ε or else .5 ≤ m ≤ 1

Re: Oxford Undergrad CS Admission Interview Problems

#27

Post by andrew1 »

sachinag wrote: post #21
andrew1 wrote: post #13

Earlier quoted context omitted.

Only some UK universities have a collegiate system (Oxford, Cambridge, Durham, St. Andrews(?)). Most do not (I believe) and I imagine are roughly similar to US universities. At collegiate universities there is a split between what a college is responsible for, and what the university is responsible for. Lectures are university-wide, exams also are (you get your degree from the university not the college), departments…

OK, this really helps; thanks! Let's say that you really wanted to study CS at Oxford; if you pick the wrong college to apply to, are you shut out from the university entirely? Or can you apply to multiple colleges at the same university at once?

Nope, you can only apply to one college at either Oxford or Cambridge per year (you can't apply to both universities, which probably makes sense as almost everyone who applied to one would probably apply to the other). There is a clearing system I think where if the college you apply to doesn't want you then other colleges can look at your application and choose to interview you, but I think that's fairly rare.

Re: Oxford Undergrad CS Admission Interview Problems

#28

Post by mjac »

ig1 wrote: post #4
apl wrote: post #3

Earlier quoted context omitted.

One more thing to keep in mind: Cambridge and Oxford interviews are more about thought process than actual solutions. I've seen candidates with correct answers not get in while peers who never found an adequate solution make it. Interesting questions, though.

I should hope that's true of interviews everywhere. It's worth noting as well that Oxford is by far the most theoretical of all the CS departments in the UK.

As a student at Cambridge, I doubt Oxford is "far" more theoretical. The London Universities like Imperial are also notable for focusing on theory over practice.

I think they assume we can learn practical skills on-the-fly and they are not as beneficial as a solid theoretical grounding.

These Oxford interview questions are more difficult than I expected, especially under Oxbridge interview conditions. Like mentioned above, they want to see how you think over what you already know.

Re: Oxford Undergrad CS Admission Interview Problems

#29

Post by ig1 »

sachinag wrote: post #21
andrew1 wrote: post #13

Earlier quoted context omitted.

Only some UK universities have a collegiate system (Oxford, Cambridge, Durham, St. Andrews(?)). Most do not (I believe) and I imagine are roughly similar to US universities. At collegiate universities there is a split between what a college is responsible for, and what the university is responsible for. Lectures are university-wide, exams also are (you get your degree from the university not the college), departments…

OK, this really helps; thanks! Let's say that you really wanted to study CS at Oxford; if you pick the wrong college to apply to, are you shut out from the university entirely? Or can you apply to multiple colleges at the same university at once?

If a college thinks you're good at your subject but doesn't want to accept you for other reasons (they have too many candidates, don't think you'd be a good fit) they can "pool" you, which means that other colleges that are interested can interview you.

You can also make an "open application" which is similar to pooling where you say you don't mind which college you get into.

Generally applicants tend to put too much emphasis on the "right" college, the academic differences aren't that great, and a lot of it is just candidates wanting to get into the more prestigious colleges for prestige's sake.

Re: Oxford Undergrad CS Admission Interview Problems

#30

Post by viraptor »

jacobolus wrote: post #25
viraptor wrote: post #9

I really like those questions - they seem very tricky until you think of some way to do it in a smaller subset or try to figure out the solution by doing the actual actions in your head. I think they're very different from the typical hiring interview logic puzzles you'd see in the companies, but I can't really express the difference... Then again, I can't really figure out (an optimal) solution to the second one...…

Question #2 also doesn’t clearly state whether we learn the value of each guess before making the next guess. As for your idea, 1/3^5 accuracy is not correct. You need to think a bit more carefully about what you can actually learn from testing a pair of points.

Ah - thanks for catching that stupid mistake :) Writing to fast and confused myself - I was thinking about ½, ½+epsilon - which is 3 segments, but not equal of course - and it went downhill from there. But yeah - I basically wanted f' at some point. More or less ½^5 accuracy.

Nice observation about learning the value before making another choice too.

Post reply on HN