I wouldn’t be surprised if what happened in underground music scenes will happen in the visual arts: a lot of pioneering electronic music (eg techno & house) exists completely off of the internet and is only available on vinyl (or on digital if you happen to be extremely lucky and know the producer). A lot of underground producers exclusively release on vinyl because they want their music to reach a smaller audience, remain in club spaces and collected by thoughtful curators. If generative ML requires data to catch on to new trends, then it seems reasonable that many visual artists might try copy the path of underground music producers and keep all their work completely offline.
In particular, for the parametric estimation setting, it can be shown that the Bayes Estimator under L_0 loss corresponds to simply finding the posterior distribution of a parameter given data, and then finding the mode of this distribution. Similarly, for L_1 loss, all we need do is find the median of the posterior distribution. And under L_2 loss, it’s just the expectation of the posterior. CMU’s 705 course is a great intro to statistical decision theory and stats more broadly for anyone interested!
(Disclaimer: I am a CMU PhD student in the machine learning department so I am somewhat biased to thinking these notes are good having taken this course myself)
If anyone is interested in topics like this and wants to delve some more into the mathematics, I’d highly recommend Tim Roughgarden’s lecture series on Algortihmic Game Theory:
That doesn’t show a graphical element of when the reminder is scheduled to occur at, as far as I can tell. But that is the type of distinction I’m thinking of: event vs reminder
Some calendars (google, native mac) allow for events that have 0 time, but they are displayed as a block of time on the calendar instead of a single horizontal line (when I look at the thin block, my OCD kicks in and asks whether the deadline is the start of the block, the end, the middle...?)
Makes me feel more and more that pen and paper is the way to go.
I was looking at the results of going from CMU to my little secondary school in Dublin, Ireland. I saw the results and saw that the last page before my Irish school was "College" and assumed it must be wrong, because how could my tiny secondary school be on the Wikipedia page for "College"? But alas, I was wrong!! I just checked and turns out it IS on the college wiki page!
I also assumed you were looking at outgoing links for both X and Y - that explains a lot.
I am super interested in this, but I have never done any graph theory or searching/planning (I'm EE) - how did you build up all of the incoming links for each wiki page? Are you storing all of this? How much data is that? Thanks for the reply!
Love the idea, and it’s brilliantly executed! Well done.
Perhaps I misinterpreted the concept of “degrees of separation”, but I was expecting the site to tell me how to start at page X and get to page Y with the min number of clicks. If you wanted to achieve this, it doesn’t strike me as appropriate to use Bidirectional BFS but IANAL.
I did notice that someone pointed out that they get different results by swapping the order of X and Y. This seems pretty surprising?
Any chance you could link the specific paper you’re talking about? This seems like a area I’d love to explore and the older you mentioned seems like a good starting point - I looked up the author on arxiv and there seems to be a few papers you could be referring to
These are all interesting questions! I would like to mention that it is important to remember when analyzing these scenarios that the world in which we live is very much a real-time system; seldom is it necessary to make decisions for "long time horizons" without the opportunity to adapt, change or update the decisions along the way.
For example, consider the hypothetical scenario where a cure for Malaria is spontaneously discovered and rolled out in Africa. Suppose, then, that in the few years following the end of Malaria, African population economists argue that an explosion is in play, to the point where the country cannot hope to continue to support itself into the future. In this scenario, I find it hard to imagine that even the most austere and strictly enforced population control program could not match the rate at which the population would increase due to the elimination of Malaria. Moreover, I would not expect that even in the hypothesized scenario, such austere measures would need to be enforced, although I will leave that to the economists and public health experts to figure out. I would suspect that most people would prefer a reality in which Malaria does not exist, but every family in Africa is only legally permitted to have one child, than a reality in which no restrictions on family size are imposed, but Malaria remains a top killer.
The point I am making here is that this is very much a real-time system, in the sense that action can be taken as soon as it is suspected that negative effects may be growing. If such negative consequences are to arise, we are not committed to a downward spiral without intervention. Furthermore, I would consider it unlikely that the rate at which the population would grow as a result of eliminating one of the leading causes of death is so great such that no intervention by the African people can keep the "net good" of the scenario positive.
(Let us suspend, for the sake of argument, the thorny path of quantification of good - suppose we use QALYs, with some simplified metric for human quality of life that is defined in such a way that makes comparisons meaningful)
Certainly, it is important to make efforts to be cognizant of negative ripple effects arising from even the most altruistic of endeavors, but it is also necessary to actually make decisions. The Horizon Effect[1] is unavoidable in situations like you are describing, but the most we can do is act conscientiously and avoid paralysis due to the inherent uncertainty that arises from our actions. I know this isn't the exact point you were raising, but reading in between your comment lies what I suspect is a hyper awareness of consequence, and I wanted to share my thoughts on that.
In response to the specific fear you mention, while this may be possible, I would consider it more important to successfully eliminate Malaria than to not do so for fear of the negative repercussions you raise. I think it is both more uncertain and more unlikely for those negative consequences to materialize in such a way that leaves humanity and leaders unable to respond adequately.
More of a motivational book than anything else, but Salman Khan's "One World Schoolhouse" was a very enjoyable book on the future of education (on and offline). It's more of a pop-sci book but it does give encouraging accounts of the success of the mastery learning philosophy for middle school math as well as Sal's vision for the future of education where institutions embrace MOOCs into their pedagogy.
I think I am in quite a similar position to you: after secondary school here in Ireland, I didn't consider any universities outside of my home town (pretty much because I didn't know a single person who was considering bigger and better options, so it genuinely didn't cross my mind to apply to Stanford, CMU, MIT etc.) so I ended up going to a fairly average and not very well known university to study electronic engineering for my undergrad.
I went on an exchange for a year to UCLA and this was when I started to feel something similar to the sentiment you're expressing here.
I'm now in my 3rd year of undergrad EE and for the last year I've been trying to fast track myself into the AI / ML field as I've been increasingly regretting my EE major and becoming more and more interested and passionate about ML (particularly the intersection of ML, altruism and design): I got Norvig & Russell's textbook and read it in outside of my engineering classes, read less technical books like Nick Bostrom's Superintelligence for motivation / food for thought, made a simple collaborative filtering recommender system using the movielens open source dataset, moved away from the web dev stuff I'd been doing in 1st and 2nd year and tried to hone in on improving my algorithm and pure CS skills, watched a load of AI / ML videos to try and get a better sense of who's who, where's where and what's going on etc. in the field. The "dream" (I use that word loosely) is to do the google brain residency program instead of a PhD, or the U Chicago data science for social good fellowship, so I've been trying to figure out how to get myself into good shape for either of them.
It's been overwhelming at times, largely because I feel like 1) I'm not in the "right" major, 2) I've had a taste of but no longer "go to" UCLA (or an equivalent high ranking university) and won't be graduating from there so will need to work hard to stand out against the competition for placements / fellowships / internships 3) I don't have mentors or peers who can help me navigate the field (I have a great relationship with a lot of my engineering professors but again, it's not ML). So I'm sort of trying to make sense of it all myself. It's reassuring to hear there are others feeling similarly and it's great to hear all that you're doing!
On a positive note, I suspect you may be overestimating the educational superiority of the top tier schools (I know I certainly did before I went to UCLA) but at the same time I don't think it's fair to completely disregard the big unis and just say "circuit theory is circuit theory" and forget about it. While I was there, I really didn't notice all that much of a difference in terms of course content or even teaching quality - the biggest difference was there were an awful lot more high achiever students in my EE classes compared to in my home uni in Ireland, and there was a much more impressive "career fair" and internship opportunity scene than at home (think Irish Cement vs Hyperloop One).
You seem to be doing everything right. I think I was edging down a "burnout" path a couple of months ago with fretting over what you're saying and over my own EE vs CS major "challenge". I've tried to take a step back and remember that there's no one enforcing a particular pace or path for me, hopefully you won't let the fretting get in the way of your passion which almost happened to me.
Just wanted to comment this to warn you about the burnout thing, reassure you somewhat about top schools and throw in a few links you might find interesting for good measure!
You mightn't find any of these links below helpful, you very well may be much more well read than myself but I thought I'd link these here anyway. The first is a reassuring AMA on reddit from the google brain team (particularly the comments where the team talk about all the different backgrounds everyone has at google brain). The second is a list of programmes, fellowships, resources and random AI / ML related pages I've encountered in the last year (amongst a lot of other stuff ). The third is a playlist I made for a friend on interesting AI / ML videos which you most likely will have seen before but you might just enjoy anyway. The quick interviews are cool if you haven't seen them already.
You could extend that argument about aesthetic design to any application though, but I think that may be framing it in the wrong way. It's (usually) not much extra effort to, say, choose a nice font or use a pleasing colour scheme. I don't think good aesthetic design should be reserved for the masses or recreational media. Good aesthetic design isn't "beneath" anything; I don't see a reason to justify bad aesthetic design in educational or instructional resources.
This reminds me strongly of when cigarette companies started to produce nicotine patches. Without sounding too cynical (it does look like a product that could be useful for people who just happen to be in noisey environments a lot) , I can't help but think Bose are trying to position themselves at the top of a new product line: hearing aids for people who damaged their own hearing through excessively loud earphone music but rebranded to not be considered so obviously as "hearing aids" in the current sense. Not that that's necessary all that bad, it makes sense for a top consumer audio company to tackle this "problem" if it is a thing; I'm by no means an ear expert but I wouldn't be surprised if a surge in hearing aid demand is due as the effects of prolonged earphone usage starts to surface.