Awesome computer science research videos
March 5, 2012 10:50 AM Subscribe
Help me find awesome videos for high school students that show interesting new algorithms research.
I'm not looking for detailed explanations of algorithms. This is to catch student interest by showing an awesome video and saying: This is real research that people in computer science are doing right now.
Two random examples: Arbitrary object tracking, and data-driven enhancement of facial attractivness
I'm not looking for detailed explanations of algorithms. This is to catch student interest by showing an awesome video and saying: This is real research that people in computer science are doing right now.
Two random examples: Arbitrary object tracking, and data-driven enhancement of facial attractivness
This TED talk discusses minimizing 'snap', or the fourth derivative of position vs. time, in order to smoothly move a flying robot, and possibly hunt humans for sport.
posted by monkeymadness at 11:21 AM on March 5, 2012 [1 favorite]
posted by monkeymadness at 11:21 AM on March 5, 2012 [1 favorite]
How about all of the quadrotor maneuvering and flocking videos? Or the google self-driving car? I've read papers from Sebastian Thrun on algorithms they've developed for matching a point cloud of objects around the car from one point in time to the next, to keep a 3D model of the world around it.
posted by pombe at 11:49 AM on March 5, 2012
posted by pombe at 11:49 AM on March 5, 2012
Best answer: Not a video, perhaps you can find one on the subject but this research where they can take a picture (even from hundreds of feet away) of your keys and, via an algorithm, produce a copy that will let them into your house is pretty scary/amazing/awesome. Especially when you think about how many pictures (potentially with keys in them) are on facebook profiles (which potentially have identifying details like addresses).
Inspire and warn, a double feature!
posted by RolandOfEld at 11:55 AM on March 5, 2012
Inspire and warn, a double feature!
posted by RolandOfEld at 11:55 AM on March 5, 2012
Best answer: The skills needed to create a video animating and teaching an algorithm is typically not acquired in a CS PhD program. Hence, research is often presented in paper form only, and sometimes source code. At most you might find someone stand at a podium using powerpoint talking to other computer scientists. Academics typically aren't paid to promote their findings directly to society, so they focus on other things.
There is one loophole in this: Computer Graphics. Since these people work with imagery and animation, they tend to produce video as a result of their research. The major journal for computer graphics is probably ACM:SIGGRAPH. SIGGRAPH regularly hits the blue, and this dude has attempted to collect a bunch of conference proceedings materials into one usable site.
Just as a recommendation, Seam Carving is a pretty neat demo, that illustrates both results, and explains how it's done. As a bonus, it's a fairly simple algorithm if dynamic programming is on the syllabus. It got picked up pretty quickly by the community; within a month someone had written a plugin for GIMP, although they neglected to fully read the paper. It might be instructive to have students read the paper, try the plugin, and ask them why resizing larger doesn't work and how it should be fixed (assuming it hasn't yet).
posted by pwnguin at 12:16 PM on March 5, 2012 [1 favorite]
There is one loophole in this: Computer Graphics. Since these people work with imagery and animation, they tend to produce video as a result of their research. The major journal for computer graphics is probably ACM:SIGGRAPH. SIGGRAPH regularly hits the blue, and this dude has attempted to collect a bunch of conference proceedings materials into one usable site.
Just as a recommendation, Seam Carving is a pretty neat demo, that illustrates both results, and explains how it's done. As a bonus, it's a fairly simple algorithm if dynamic programming is on the syllabus. It got picked up pretty quickly by the community; within a month someone had written a plugin for GIMP, although they neglected to fully read the paper. It might be instructive to have students read the paper, try the plugin, and ask them why resizing larger doesn't work and how it should be fixed (assuming it hasn't yet).
posted by pwnguin at 12:16 PM on March 5, 2012 [1 favorite]
Best answer: Pathfinding might be a good example. In video games, for example, the user might tell a unit to move to a given location, and the unit has to find a good path to get there. What you need to do this is an algorithm that both finds a path that's short (ideally as short as possible), while still running fast enough that it can be computed in close to real time. It's a good application of practical AI.
Nathan Sturtevant (disclaimer: we used to be at the same University), a professor at the University of Denver, has some videos on his website using maps from Dragon Age as a research testbed, and shows how several different search algorithms search through the space in order to find a good route to their destination:
http://movingai.com/flrta.html
http://movingai.com/astar.html
http://movingai.com/astar-var.html
I've seen him give presentations a few times where he had a pretty big set of similar videos. If you asked him, he might have more than what's online that he'd be able to share.
posted by Fully Completely at 12:16 PM on March 5, 2012
Nathan Sturtevant (disclaimer: we used to be at the same University), a professor at the University of Denver, has some videos on his website using maps from Dragon Age as a research testbed, and shows how several different search algorithms search through the space in order to find a good route to their destination:
http://movingai.com/flrta.html
http://movingai.com/astar.html
http://movingai.com/astar-var.html
I've seen him give presentations a few times where he had a pretty big set of similar videos. If you asked him, he might have more than what's online that he'd be able to share.
posted by Fully Completely at 12:16 PM on March 5, 2012
Oh bother, "SIGGRAPH regularly hits the blue" was supposed to link to Mefi's siggraph tag.
posted by pwnguin at 12:25 PM on March 5, 2012
posted by pwnguin at 12:25 PM on March 5, 2012
The Beautiful Math behind the World's Ugliest Music.
posted by spinifex23 at 12:29 PM on March 5, 2012
posted by spinifex23 at 12:29 PM on March 5, 2012
NSF's video team has put together many videos exploring different aspects of science and engineering. The Profiles of Scientists and Engineers series (here's one on virtual reality) seems a great first place to look, since they are made to show students possible careers and a "day in the life" of a scientist. (They're just like you and me! Really!) After that I'd search their multimedia gallery or specifically check out some of their video features series on this page. A quick search turned up this Innovators video on recent robotics work. Besides that series I'd recommend looking at the "Science Nation" episodes which run about 3" long I think, and cover a variety of subjects.
My disclaimer is that I used to work there and therefore may be biased, (and hope this doesn't count as a self-link!), but I do think they're worth considering. They're on YouTube too.
posted by NikitaNikita at 4:32 PM on March 5, 2012
My disclaimer is that I used to work there and therefore may be biased, (and hope this doesn't count as a self-link!), but I do think they're worth considering. They're on YouTube too.
posted by NikitaNikita at 4:32 PM on March 5, 2012
Oh, and here's another possible source of good videos. AAAI, the Association for the Advancement of Artificial Intelligence, runs one of the top-tier conferences in the field of artificial intelligence. Since 2007, they've had a video competition that goes along with the annual conference. I've only seen a few of the videos at the conference, but they might be close to what you're looking for:
http://videolectures.net/aaai_video_competition/
Also, Dr. Andrew Ng at Stanford is famous in AI / Reinforcement Learning circles for having awesome videos to demonstrate what he's working on. In particular, his autonomous helicopter videos are great:
http://heli.stanford.edu/
And autonomous powerslide parallel parking, too:
http://spectrum.ieee.org/automaton/green-tech/advanced-cars/autonomous-car-learns-to-powerlslide-into-parking-spot
Really, just about any video that he's associated with is probably going to be what you're looking for.
I'll see if I can turn up a few more good sources for short, exciting videos.
posted by Fully Completely at 8:52 PM on March 5, 2012
http://videolectures.net/aaai_video_competition/
Also, Dr. Andrew Ng at Stanford is famous in AI / Reinforcement Learning circles for having awesome videos to demonstrate what he's working on. In particular, his autonomous helicopter videos are great:
http://heli.stanford.edu/
And autonomous powerslide parallel parking, too:
http://spectrum.ieee.org/automaton/green-tech/advanced-cars/autonomous-car-learns-to-powerlslide-into-parking-spot
Really, just about any video that he's associated with is probably going to be what you're looking for.
I'll see if I can turn up a few more good sources for short, exciting videos.
posted by Fully Completely at 8:52 PM on March 5, 2012
« Older A new transmission? Really? It's a Toyota? | Can one man still create a complex website Newer »
This thread is closed to new comments.
posted by empath at 11:11 AM on March 5, 2012