A Faster Internet - Designed By Computers?

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
Researchers from MIT’s Computer Science and Artificial Intelligence Laboratory claim to have developed computer designed algorithms that significantly outperformed algorithms devised by human engineers.

In tests that simulated a high-speed, wired network with consistent transmission rates across physical links, Remy’s algorithms roughly doubled network throughput when compared to Compound TCP and TCP Cubic, while reducing delay by two-thirds. In another set of tests, which simulated Verizon’s cellular data network, the gains were smaller but still significant: a 20 to 30 percent improvement in throughput, and a 25 to 40 percent reduction in delay.
 
Guess we'll be out of a job sooner than we thought. :(
We'll need you in the future when we have to disable and destroy them before they destroy of of us. ;)


You know there will be some maniac that uses AI in a robot that replicates itself and destroys humans. This will happen, I've seen in in the movies.
 
Here is the distinction. These algorithms can be put into play immediately without worry of patent because a computer devised them and those computer aren't going to sue. Yet.
 
Here is the distinction. These algorithms can be put into play immediately without worry of patent because a computer devised them and those computer aren't going to sue. Yet.

My computer sued me for spending most of my time at icanhascheezburger.com :(
 
wonder if they used a "genetic algorthym" which is basically a brute force design method using some sort of evolution method to get a "better" outcome.
 
For this to work, they have to teach computers the inarguable importance of porn.... that way it gets it's due amount of resources
 
Just a matter of time when the algorithm spawns another algorithm that burns MIT to the ground


Posted from Hardforum.com App for Android
 
I like it. I wonder if the increased routing efficiency will increase or reduce overall utilization of core routers. My guess is it increases it, which will result in greater power usage, increased cooling costs and reduced equipment longevity. This is akin to supercharging an engine, and that performance doesn't come without a cost.
 
We must take The Grid to its maximum efficiency.

What efficiency? Talk to a facilities manager and they will tell you efficiency is getting the lowest power bill. Talk to a geek and they will tell you it's all about overclocking with total disregard to power consumption.
 
Sounds very similar to applying a genetic algorithm to a problem. They will get a solution that works well / better, but then they won't understand exactly why it is better. NASA did this when searching for an optimum antenna design.

http://idesign.ucsc.edu/projects/evo_antenna.html

Cool - I've always thought a brute force algorithm approach would be the end goal of good AI; there are so many applications waiting for something like this to happen to them:
* Car engines
* Rockets
* Energy transferal
* Medicine / surgery

Similar to how a computer can examine a chess game and play better than 99.99% of the people on the planet, once the programming is there and sufficiently advanced, we should be able to solve a lot of our engineering problems programmatically. Everything from the best-possible bridge design to cars that are way more efficient, batteries which last longer, etc..

I foresee a "new wave" of engineering in the future, designed via AI brute forcing the problem to come up with the "best possible" way of doing something.
 
Cool - I've always thought a brute force algorithm approach would be the end goal of good AI
+1 very cool... except I wouldn't use the term "brute force"; that term is reserved for use in describing Haggar from Final Fight
 
Cool - I've always thought a brute force algorithm approach would be the end goal of good AI; there are so many applications waiting for something like this to happen to them:
* Car engines
* Rockets
* Energy transferal
* Medicine / surgery

Similar to how a computer can examine a chess game and play better than 99.99% of the people on the planet, once the programming is there and sufficiently advanced, we should be able to solve a lot of our engineering problems programmatically. Everything from the best-possible bridge design to cars that are way more efficient, batteries which last longer, etc..

I foresee a "new wave" of engineering in the future, designed via AI brute forcing the problem to come up with the "best possible" way of doing something.

I'm not sure that AI is the correct term to describe that. Especially when talking about brute force.

All it is really doing is going by a set code path that is programmed to see results of previous runs and automatically try different variables.
 
Back
Top