Author |
Topic: Neat Ideas in Computer Science (Read 2294 times) |
|
Amir Michail
Guest
|
I am contemplating setting up a slashdot-style site for discussing neat new computer science ideas, particularly in more applied areas (e.g., web search, collaborative filtering, social networks, ESP game for labeling web images, software engineering tools, etc.). The emphasis would be on ideas that most people would understand without much technical expertise in the area. The ideas may come from recent publications or may merely be speculative discussions on possibly interesting new research directions. As an example of a speculative discussion, we have already seen a move to using people to solve hard problems such as image labeling using the ESP game. The idea is based on setting up problems as something fun that people would like to do in their spare time. And so the question arises, given some problem we would like to solve using a computer/human hybrid system, how can we set up a game that people would find fun? After all, if it is not fun, few people will play it and the algorithm would not scale. And so, it seems we have a mix of computer science and psychology/sociology. As another example of a speculative discussion of computing using people: could we use the massively parallel human visual system to solve hard combinatorial problems? Could we find an NP-complete visual problem that people could do well on? (And so by reduction, solve other NP-complete problems also?) Is there a place for discussions like these already on the web? I know about http://www.sweetcode.org, but what I have in mind would not only present new ideas but also discuss them slashdot-style. Moreover, I have a few ideas of my own to offer in multiple fields for which I would appreciate feedback. For example, see http://www.cse.unsw.edu.au/~drt/screenshot and http://opsis.sourceforge.net/screenshots.html. Of course, if there is interest here, perhaps a new category could be added here for this purpose. Amir
|
|
IP Logged |
|
|
|
towr
wu::riddles Moderator Uberpuzzler
Some people are average, some are just mean.
Gender:
Posts: 13730
|
|
Re: Neat Ideas in Computer Science
« Reply #1 on: Dec 22nd, 2003, 4:44am » |
Quote Modify
|
on Dec 21st, 2003, 4:31pm, Amir Michail wrote:Could we find an NP-complete visual problem that people could do well on? (And so by reduction, solve other NP-complete problems also?) |
| Only if they're small problems. The problem is that our parallel system is relatively small compared to some problems.. I would guess the number of parallel nodes would have to be in the order of the size of the input. Besides, parallel computing works best when divide and conquer approaches work (and thus each node can work relatively independant). And those kinds problems can generally be solved in polynomial time, I think..
|
|
IP Logged |
Wikipedia, Google, Mathworld, Integer sequence DB
|
|
|
Amir Michail
Guest
|
|
Re: Neat Ideas in Computer Science
« Reply #2 on: Dec 25th, 2003, 7:24pm » |
Quote Modify
Remove
|
on Dec 21st, 2003, 4:31pm, Amir Michail wrote:I am contemplating setting up a slashdot-style site for discussing neat new computer science ideas, particularly in more applied areas (e.g., web search, collaborative filtering, social networks, ESP game for labeling web images, software engineering tools, etc.). |
| I have decided to start up such a site: http://www.cse.unsw.edu.au/~papers/PostNuke/html Although it is currently called a "UNSW CSE Forum", contributions from all are welcome. Amir
|
|
IP Logged |
|
|
|
|