[00:44:03] *** Quits: r0bby_ (~wakawaka@guifications/user/r0bby) (Ping timeout: 276 seconds) [00:52:43] *** Joins: SamuelKelemen (~DarkNova@c-68-35-150-201.hsd1.al.comcast.net) [00:57:27] *** Quits: SamuelKelemen (~DarkNova@c-68-35-150-201.hsd1.al.comcast.net) (Ping timeout: 264 seconds) [01:42:48] *** Joins: munja (~munja@wf121-187.ust.hk) [02:48:54] *** Quits: munja (~munja@wf121-187.ust.hk) (Quit: munja) [04:44:38] *** Joins: munja (~munja@wf121-187.ust.hk) [07:18:20] *** Quits: munja (~munja@wf121-187.ust.hk) (Quit: munja) [08:32:15] *** Joins: munja (~munja@wf121-187.ust.hk) [09:22:07] *** Quits: lablayers (uid55894@gateway/web/irccloud.com/x-ephfdlokayafcayj) (*.net *.split) [09:24:02] *** Joins: lablayers (uid55894@gateway/web/irccloud.com/x-ephfdlokayafcayj) [09:26:54] *** Quits: iangcarroll (uid74403@gateway/web/irccloud.com/x-nmcwxmudrzvwrszz) (Ping timeout: 248 seconds) [09:34:59] *** Joins: iangcarroll (uid74403@gateway/web/irccloud.com/x-kksyhdpygtzpqeyn) [09:40:58] *** Quits: lablayers (uid55894@gateway/web/irccloud.com/x-ephfdlokayafcayj) (Quit: Connection closed for inactivity) [11:42:59] *** Quits: munja (~munja@wf121-187.ust.hk) (Quit: munja) [17:42:00] http://i.imgur.com/Y596Sjs.png [17:42:05] ultimate rekage right here [17:42:25] the desperation is real…because the real hackers are going to bitcamp [17:42:31] yikes can anybody here help me out with some machine learning stuff? [17:42:42] OMG vishwin I got off the bitcamp waitlist [17:42:46] I didn't know i was ON a waitlist [17:42:54] :-) [17:43:03] so I won't be going since I made other commitments lol [17:43:11] damn [17:43:23] I mean, my hackathon season has been done since brickhack (cold_sauce) [17:43:31] I've never actually been to a hackathon :p [17:43:38] but penn state's hackathon is a total joke [17:44:04] this is just one exhibit of how joke it is [17:44:09] lol [17:44:15] kinda like mhacks this semester eh [17:44:19] oh yeah. [17:44:22] definitely. [17:44:32] but even worse because these organisers don't actually hack [17:44:38] especially not with people they don't know [17:44:55] they're good with business logic and not much else [17:45:02] lol [17:46:01] speaking of MHacks, the organisers' incompetence is the main reason why there's apparently still a hot tub at shift. [17:48:18] lmao [17:48:23] so vishwin do you know any machine learning? :) [17:49:27] unfortunately not :-( [17:49:37] not my area of expertise/interest :-\ [17:49:48] well crap [17:49:54] i have no idea where even to ask [19:28:19] clinth: are you intrested in learning about it? [19:28:25] I'm taking a course right now on Coursera [19:28:27] it just started [19:28:31] yeah I am interested [19:28:33] on Machine Learning by Andrew Ng [19:28:39] it's amazing [19:28:43] oh yes I've seen a lot of this [19:28:43] it's the course that started Coursera [19:28:46] but it kind of... [19:28:50] it's very math-light :p [19:28:56] really? [19:29:08] a friend of mine who took it said that he thinks it's way too "theoretical" [19:29:24] oh no I think it's not at all heavy on theory [19:29:33] interesting [19:30:55] yeah I dunno [19:30:57] I didn't finish it though [19:30:59] I only got to week 6 [19:31:02] well nevermind [19:31:05] just talked to him [19:31:11] turns out he just did the half that was theoretical [19:31:16] :p [19:31:18] oops [19:31:20] oh is it the second half? [19:31:58] hmm I'm not sure [19:32:00] let me ask [19:35:09] yeah [19:35:12] that's what he says [19:35:31] huh interesting [19:35:36] because the second half to me looks more applied [19:35:47] probably changed since he took it maybe [19:35:54] it's possible [19:35:55] i dunno [19:35:59] it's interesting I guess but [19:36:07] I'll take the intro ML course next semester [19:36:31] every single example that he provides [19:36:34] is just mind blowing to me [19:36:51] I've never taken a class where I feel that way until now [19:36:59] lol like what? [19:37:37] like having one microphone close by one person and another one close to a second person [19:37:51] and then the algorithm can single out a conversation [19:38:10] even in a loud room [19:38:15] *** Joins: baristaTam (~baristata@95.85.34.133) [19:38:29] yea it's neat but [19:38:32] i want to know more about the math behind it [19:38:37] not just like "this is something cool it can do" [19:38:42] and that's where the course seemed to be lacking [19:39:03] maybe i should take a look at the actual lectures from like andrew ng's course at stanford [19:39:10] where he teaches with a blackboard and stuff instead of on coursera :p [19:39:47] well I think he's first introducing the concepts [19:39:53] have you gone further in it? [19:39:58] because it seems like we get into that sttuff [19:40:01] later on [19:40:49] in the FAQ it talks about how they use octave instead of R because they want you to implement the math behind the algos [19:40:51] instead of using R [19:40:56] which just has the algos set up for you [19:41:08] which makes me feel like they care about teaching the math behind it [19:43:04] oh interesting [20:00:17] if you want the math heavy version of that course [20:00:27] watch andrew ng's lectures on youtube [20:00:35] i stopped going to my ML class in college [20:00:37] and just did that [20:00:54] my prof's supervisor in college was Andrew Ng [20:01:03] so the course was pretty much in the same format [20:02:03] oh [20:02:07] i'm taking 445 next semester [20:02:14] which prof? [20:02:18] if it's honglak [20:02:20] it's not worth it [20:02:21] abernethy and deng [20:02:26] you should be fine then [20:02:29] lol [20:02:32] but still watch the youtube lectures [20:02:34] people in MSAIL said honglak is good? [20:02:34] they're great [20:02:38] lolol [20:02:40] no [20:02:44] worst professor i've had [20:02:46] ever [20:02:48] actually [20:02:52] maybe sugi in 489 was worse [20:03:04] i have only ever heard one good thing said about 489 [20:03:11] and it was from an alum engineer at the career fair [20:03:26] i have a writeup about why i hate that course in the CSE group [20:03:27] i think i will just...not take it and hope i can get a job anyway :p [20:03:44] oh yeah I think I've read that haha [20:03:48] lol [20:04:04] do you have any favorite ULCS? [20:06:55] i liked compilers because i've always been interested in PL [20:07:08] i was going to take that this semester until it disappeared from the schedule... [20:07:14] lol [20:07:16] it's why i'm in 388 lol [20:07:22] 388 was a favorite of mine [20:07:24] but i sorta like 388 so its fine [20:07:24] obviously [20:07:46] yea it's interesting enough I guess [20:07:50] I just don't feel like it's really taught me anything [20:08:00] CTFs man [20:08:01] but that's no fault of the course [20:08:04] that's where you'll learn everything [20:08:09] i regretted not taking 475 for a while [20:08:14] i took 475 before 388 [20:08:22] but then i learned way more from CTFs than i would have ever learned from that course [20:08:37] lol [20:08:54] I'm not sure that my team got any of the crypto ones last weekend though :p [20:09:11] all the git ones were my favorites :D [20:09:20] i usually spend a few weeks to months on problems after a CTF [20:09:21] so [20:09:25] ankit wrote the first git one [20:09:30] and i was like, this is way to easy [20:09:33] so i wrote two more [20:09:35] one of them on the plane [20:09:36] which was the first one? [20:09:37] lol [20:09:47] you could solve it with git reflog [20:09:48] it was the rebase one? [20:09:50] yea [20:09:59] yea i used git reflog for that [20:09:59] probably other ways to solve it as well [20:10:47] but yea, there are certain problems i come back to a year after the CTF [20:10:50] and try to do them [20:10:58] that's how i got good at crypto [20:11:22] lol [20:11:23] if there were problems you didn't solve [20:11:28] you should keep trying [20:11:33] there were a lot of those :p [20:11:50] that means you have a lot to learn [20:11:51] we were second place, but we were still like 1500 points or something crazy behind first place [20:12:34] how much is it supposed to be a team effort? [20:12:48] it depends [20:12:54] at the beginner level, it's very much a team effort [20:13:06] at harder levels, it still is, but not as much [20:13:23] at harder levels, you really need to solve things fast [20:13:32] and putting all resources on a single problem is wasteful [20:13:33] if you don't need it [20:13:35] imo having people working on different problems simultaneously would have been helpful points-wise [20:13:51] yea, really depends on your team [20:14:01] my team was kind of... me :( [20:14:05] lol [20:14:09] you should go to CTF club [20:14:20] pretty sure all the people on the winning team were in it [20:14:26] interesting lol [20:14:43] i think the other people on my team got maybe... 35 points themselves? :\ [20:14:49] the apple one, and like a quiz lol [20:14:56] lol [20:15:36] i mean i still learned a lot and had fun, so that's what is important [20:17:12] someone posted on piazza afterwards like [20:17:39] mfw when i walk in and see kegan is my opponent: http://vignette3.wikia.nocookie.net/familyguy/images/9/99/Stewie_gun_mouth_super.jpg/revision/latest?cb=20131211011839 [20:26:50] *** Quits: Monie (~Monie@cowbot.jamonek.com) (Remote host closed the connection) [22:08:01] our best classifier is just fucking naive bayes on bigrams [22:08:02] great [22:46:33] what's it's recognition rate? [22:47:19] also, why not consider a nn or svm approach? [22:47:42] the math is vectory enough that a gpu approach wouldn't be too bad, and would help when you have enough hidden layers and intermediate nodes [22:50:07] around 71% m0shbear [22:50:16] we're considering trying an svm [22:50:26] none of us really know anything about ML though, which is the problem [22:51:04] doing naive bayes with *emojis* for male/female actually is at like 66% lmao [23:52:10] *** Joins: robbyoconnor (~wakawaka@guifications/user/r0bby) [23:52:10] *** Quits: robbyoconnor (~wakawaka@guifications/user/r0bby) (Client Quit)