Re:Badass compression algorithm? -- NOT
on
Share The Pi!
·
· Score: 1
Unfortunately, you need to specify the index to the beginning of the message. Since your message is probably a very long way into Pi, the index will likely be more bits than the message itself. I discovered this property of compression when I tried to build a compression routine based on Godel numbers. Most compression algorithms use the assumption that there are repeatable elements that can be compressed. This is sort of like using fewer bits to represent the low frequency component of a signal. If the message doesn't have much repetition, that algorithm will do poorly. You could also write a compression algorithm that knew commonly used phrases - this algorithm will only work well on phrases in the domain for which it was designed. All of this is related to Wolpert's No Free Lunch theorem.
I've seen two arcade games that have the same effect. One is a light gun game where you actually have to duck behind things. The other is a boxing game where you throw punches and duck the other guys punches. I've only played the cop game, and I thought it was pretty fun (and tiring). The screen perspective changes as you move (I think Caramack has mentioned he would like to see this). Imagine Quake (or any FPS) with this device. You could still use a game pad for gross movement, but you could use your entire body for fine aiming and dodging.
Actually, I remember him saying that if he wanted more money he would make another Star Wars sequel (which he did -- and apparently with more interest in making money than a great addition to the saga)
Re:Good movie- Bad ending(SPOILER)...
on
Review: A.I.
·
· Score: 1
They weren't aliens. They were the highly advanced AI's that were left after everybody else had gone. They were searching to understand their creator -- much like we often do.
The last screen of the trailer has a series of dots that decode to the following phone number ( I tried to put the dots in, but slashdot enacted its lameness filter against ascii art)
(503) 321 5122
It won't really matter if more people are added. To test the solution you need to post to a web server that can only handle about 4 attempts per second anyway. Currently, attempts are being made to reduce the search space by using clue available to us.
The quote refers to Basic, not VisualBasic. Basic consisted of a bunch of numbered instructions. There were no functions or parameters - not to mention classes. Just numbered labels and goto statements. The closest it had to a function was gosub - which pushed the current line onto a stack and let you return from it. My memory is a bit fuzzy, but I am pretty sure that you didn't even have local variables. Everything was global. It did not lend itself to structured programming and it took a change in thinking to switch to a language like C.
Around what percentage of the TMBG unlimited price do you guys see? Or more generally, how does the deal compare to the ones that you get from more traditional publishers?
This is the third time slashdot has posted this story. The company used to be called C3D. A flourescent effect is used to illuminate individual layers in a multilayer CD. This effect allows a laser to read only that layer.
I got to play with a Phantom during a recent tour of the AI lab at MIT. Way cool stuff. They are doing research on using this for minimally invasive surgery tecniques. Tools enter the body through a small hole and surgeons can using imaging techniques and feed back from devices like the phantom to operate without acutally see things. I think it can also be used to train surgeons since it can simulate the forces of a real surgical procedure.
There is a group at UNC using a Phantom to provide tactile feedback when manipulating viruses. These things start around $10,000 though.
It is interesting if you look at Amazon's stance on this. Slashdot pointed to an interview of Amazon's CEO a few months back. The patent is not on linking a cookie to credit card data. The patent is on the idea that you can click on any item to purchace it. The item does not go to a shopping cart for checkout later - it is purchaced when you click on it. This is an innovative idea (and a bad one IMHO - I hate the idea) that, I think, was not done before Amazon. The patent does not cover linking cookies to credit card data except as a way to implement the idea above.
Look at www.accutemp.net. The company specializes in cooking equipment. Specifically griddles and steamers. The robot was meant as a promotional bit at a trade show -- "This new griddle boasts such accurate and consistent temperature control across its surface that even a robot can cook on it." This makes me think that the robot is extremely stupid which makes the story much less interesting. I'd rather see a robot that could cook well on a bad griddle -- "This new robot is so advanced it can make soufflé a in a cast iron pot over a camp fire."
A robot arm in a manufacturing plant is one thing. The robots I have seen know exactly where the part is and where it goes. It is just following a programmed path that could almost be done with logo. A robot that makes burgers needs to be a little more complicated. It needs to be able to see the burgers and manipulate the spatula to move under them. It needs to keep track of how long each burger is cooking. It needs to be able to retrieve new burgers. I don't know how advanced this design is. Does it do these things or do they cheat on some of them (put all the burgers on at once, flip them all in 2 minutes, remove them all in 3 more minutes). How does the robot get new burgers? Do they have to be in a human filled hopper or can it recognize a box of burgers, reach in and take one?
These are called genetic algorithms and are a fairly established area of AI research. GA's often produce convoluted solutions and sometimes produce novel solutions that work better than a person might have produced.
One experiment I read about described the generation of a sorting algorithm in a distributed environment. The best solution worked very well and was very complicated. The researcher said that he was unable to describe the algorithm in terms more simple than the algorithm itself.
This reminds me of the time several years ago when the license agreement said you couldn't write an operating system with their compiler. I think the end result was some lawyer stuck it in without really understanding it. Borland seems to have a history of poorly written license agreements.
I recently took a class covering some topics like this.
Ants use pheremone trails to find fairly optimized paths to food. Some researchers have adapter their strategy to route IP packets. They use 'pheremone bits' attached to the packets headers to get information about the networks. They then use the information to route packets more efficiently. As the network properties change the routing algorithm adapts. Supposedly, it works much better than the current method (which finds the shortest path - i think ?).
I bought a Canopus DVRex and a 40GB MedeaRAID a few years ago and have been very pleased with it. I shoot on a Sony DSR200A. The video goes into the computer, I edit, and it come out on video looking very nice. It works great on a K6-2/400. It worked better when I had one of the PCI video cards they recommended, but when I switched to an ATI AGP card the video didn't keep up on the screen (the video output was still great though). I haven't had any problems with the MedeaRAID, but I do feel like I could have gotten a cheaper soluttion that works just as well. Feel free to contact me via email if you have any questions.
Thank you for the information. BTW, I've played around with evolving neural networks a little. Here is some Java source I wrote to a fully connected net that learns to count to 7 using only one input to tell it when to start counting. It is based on the technique that Foley and Chellapilla used to evolve a checkers playing network. The program is pretty simple but I found it interesting. I wrote it for a class I am currently taking.
I am planning to attend graduate school in one to two years. My interests lean heavily toward topics of evolutionary computing (and using biological models for computation in general). It looks like Brandeis has a good research group for this topic, but I've found it difficult to find other schools with groups working heavily with evolution. Do you know what other graduate schools are have research groups in these areas?
Slashdot Mirror
Unfortunately, you need to specify the index to the beginning of the message. Since your message is probably a very long way into Pi, the index will likely be more bits than the message itself. I discovered this property of compression when I tried to build a compression routine based on Godel numbers. Most compression algorithms use the assumption that there are repeatable elements that can be compressed. This is sort of like using fewer bits to represent the low frequency component of a signal. If the message doesn't have much repetition, that algorithm will do poorly. You could also write a compression algorithm that knew commonly used phrases - this algorithm will only work well on phrases in the domain for which it was designed. All of this is related to Wolpert's No Free Lunch theorem.
I've seen two arcade games that have the same effect. One is a light gun game where you actually have to duck behind things. The other is a boxing game where you throw punches and duck the other guys punches. I've only played the cop game, and I thought it was pretty fun (and tiring). The screen perspective changes as you move (I think Caramack has mentioned he would like to see this). Imagine Quake (or any FPS) with this device. You could still use a game pad for gross movement, but you could use your entire body for fine aiming and dodging.
Actually, I remember him saying that if he wanted more money he would make another Star Wars sequel (which he did -- and apparently with more interest in making money than a great addition to the saga)
They weren't aliens. They were the highly advanced AI's that were left after everybody else had gone. They were searching to understand their creator -- much like we often do.
The last screen of the trailer has a series of dots that decode to the following phone number ( I tried to put the dots in, but slashdot enacted its lameness filter against ascii art) (503) 321 5122
It won't really matter if more people are added. To test the solution you need to post to a web server that can only handle about 4 attempts per second anyway. Currently, attempts are being made to reduce the search space by using clue available to us.
The theory is that the game takes place several decades after the movie. Some characters overlap, but mostly it is a different story.
The quote refers to Basic, not VisualBasic. Basic consisted of a bunch of numbered instructions. There were no functions or parameters - not to mention classes. Just numbered labels and goto statements. The closest it had to a function was gosub - which pushed the current line onto a stack and let you return from it. My memory is a bit fuzzy, but I am pretty sure that you didn't even have local variables. Everything was global. It did not lend itself to structured programming and it took a change in thinking to switch to a language like C.
Oh goodie! My wrists are already shot. Lets start working on my neck now.
Around what percentage of the TMBG unlimited price do you guys see? Or more generally, how does the deal compare to the ones that you get from more traditional publishers?
This is the third time slashdot has posted this story. The company used to be called C3D. A flourescent effect is used to illuminate individual layers in a multilayer CD. This effect allows a laser to read only that layer.
There is a group at UNC using a Phantom to provide tactile feedback when manipulating viruses. These things start around $10,000 though.
It is interesting if you look at Amazon's stance on this. Slashdot pointed to an interview of Amazon's CEO a few months back. The patent is not on linking a cookie to credit card data. The patent is on the idea that you can click on any item to purchace it. The item does not go to a shopping cart for checkout later - it is purchaced when you click on it. This is an innovative idea (and a bad one IMHO - I hate the idea) that, I think, was not done before Amazon. The patent does not cover linking cookies to credit card data except as a way to implement the idea above.
Look at www.accutemp.net. The company specializes in cooking equipment. Specifically griddles and steamers. The robot was meant as a promotional bit at a trade show -- "This new griddle boasts such accurate and consistent temperature control across its surface that even a robot can cook on it." This makes me think that the robot is extremely stupid which makes the story much less interesting. I'd rather see a robot that could cook well on a bad griddle -- "This new robot is so advanced it can make soufflé a in a cast iron pot over a camp fire."
AccuTemp Products
A robot arm in a manufacturing plant is one thing. The robots I have seen know exactly where the part is and where it goes. It is just following a programmed path that could almost be done with logo. A robot that makes burgers needs to be a little more complicated. It needs to be able to see the burgers and manipulate the spatula to move under them. It needs to keep track of how long each burger is cooking. It needs to be able to retrieve new burgers. I don't know how advanced this design is. Does it do these things or do they cheat on some of them (put all the burgers on at once, flip them all in 2 minutes, remove them all in 3 more minutes). How does the robot get new burgers? Do they have to be in a human filled hopper or can it recognize a box of burgers, reach in and take one?
One experiment I read about described the generation of a sorting algorithm in a distributed environment. The best solution worked very well and was very complicated. The researcher said that he was unable to describe the algorithm in terms more simple than the algorithm itself.
This reminds me of the time several years ago when the license agreement said you couldn't write an operating system with their compiler. I think the end result was some lawyer stuck it in without really understanding it. Borland seems to have a history of poorly written license agreements.
The first link should be:Computational Beauty of Nature
Ants use pheremone trails to find fairly optimized paths to food. Some researchers have adapter their strategy to route IP packets. They use 'pheremone bits' attached to the packets headers to get information about the networks. They then use the information to route packets more efficiently. As the network properties change the routing algorithm adapts. Supposedly, it works much better than the current method (which finds the shortest path - i think ?).
Some links: Computational Beauty of Nature
Mobile Software Agents for Dynamic Routing
I bought a Canopus DVRex and a 40GB MedeaRAID a few years ago and have been very pleased with it. I shoot on a Sony DSR200A. The video goes into the computer, I edit, and it come out on video looking very nice. It works great on a K6-2/400. It worked better when I had one of the PCI video cards they recommended, but when I switched to an ATI AGP card the video didn't keep up on the screen (the video output was still great though). I haven't had any problems with the MedeaRAID, but I do feel like I could have gotten a cheaper soluttion that works just as well. Feel free to contact me via email if you have any questions.
Thank you for the information. BTW, I've played around with evolving neural networks a little. Here is some Java source I wrote to a fully connected net that learns to count to 7 using only one input to tell it when to start counting. It is based on the technique that Foley and Chellapilla used to evolve a checkers playing network. The program is pretty simple but I found it interesting. I wrote it for a class I am currently taking.
I am planning to attend graduate school in one to two years. My interests lean heavily toward
topics of evolutionary computing (and using biological models for computation in general).
It looks like Brandeis has a good research group for this topic, but I've found it difficult to
find other schools with groups working heavily with evolution. Do you know what other graduate
schools are have research groups in these areas?