Man vs. Machine
-
:laugh: Fair point, I'll rephrase the question: Will there come a point where the human general has to resort to extreme violence or praying that the cleaner unplugs the computer?
Whenever you find yourself on the side of the majority, it is time to pause and reflect. - Mark Twain
PeejayAdams wrote:
human general has to resort to extreme violence
Every human on earth has the propensity for evil and violence. so yes, the human general has to resort to extreme violence; it is his nature.
-
musefan wrote:
So bottom line, I think it really depends on how prepared the human is before starting, and very much how the machine is programmed to play.
Yes, I think we should go under the assumption that every human involved - programmer(s) and player alike - is as good as they can possibly be.
Whenever you find yourself on the side of the majority, it is time to pause and reflect. - Mark Twain
-
I was looking at [this problem](https://puzzling.stackexchange.com/questions/89241/the-100-soldier-problem) which is essentially a game known as [Colonel Blotto](https://en.wikipedia.org/wiki/Blotto\_game) Each general must allocate a given number of troops across a given number of platoons which then battle against each other with results being decided on numerical supremacy. To win you need to both win the greatest number of battles and have the highest number of remaining troops. So what we have is a zero-sum game which has twin objectives (somewhat like hi-lo poker variants but with a requirement to win both hands) with the additional complication of sequencing (a la rock-paper-scissors - e.g. if I set my troops out [19, 1, 19, 1, 19, 1, 19, 1, 19, 1], I might get a very different result than I would be sending out [1, 19, 1, 19, 1, 19, 1, 19, 1, 19] even though I'm using an identical strategy). All of which gets me wondering - how well could we get a machine to play this game against a human opponent? For the first element, calculating the optimal troop split, we could simply brute-force every possible combination, play them against each other and see what wins. Given the vast number of combinations involved, it's obviously a huge task but it's clearly do-able. We're essentially going to wind up with a ranked list of possible distributions). Let's assume that the results of these calculations will be revealed/knowable to the human opponent. Now the human general has an advantage, he knows that the computer knows that the best distribution is distribution X and can easily counter (let's suppose the calculated "optimal" distribution is actually 10 platoons of 10 and the human knows that the computer will play the "best" distribution, the human simply plays 9 platoons of 11 and 1 of 0 and crushes him every time). So the computer needs to throw a few curve balls - the easiest way to do this is to pick a random distribution from the top n distributions. The computer could further obfuscate by applying random increments and decrements to various pairs within the distribution or whatever. Similarly when it comes to ordering the platoons, the computer's most effective approach might just be to randomise the whole distribution array, maybe it can work a more intricate algorithm based on previous distributions employed by the human general. The human of course, has all the same curve balls at his disposal. Machine learning is definitely on the menu and time is no obj
PeejayAdams wrote:
Machine learning is definitely on the menu and time is no object.
Try an evolutionary algorithm[^]. It will produce results far more quickly than brute force and is far easier to implement than machine learning. The whole problem looks very well suited for such an algorithm.
I have lived with several Zen masters - all of them were cats. His last invention was an evil Lasagna. It didn't kill anyone, and it actually tasted pretty good.
-
PeejayAdams wrote:
human general has to resort to extreme violence
Every human on earth has the propensity for evil and violence. so yes, the human general has to resort to extreme violence; it is his nature.
Slacker007 wrote:
Every human on earth has the propensity for evil
The true question here is what kind of evil. Chaotic evil, neutral evil or lawful evil.
I have lived with several Zen masters - all of them were cats. His last invention was an evil Lasagna. It didn't kill anyone, and it actually tasted pretty good.
-
PeejayAdams wrote:
Machine learning is definitely on the menu and time is no object.
Try an evolutionary algorithm[^]. It will produce results far more quickly than brute force and is far easier to implement than machine learning. The whole problem looks very well suited for such an algorithm.
I have lived with several Zen masters - all of them were cats. His last invention was an evil Lasagna. It didn't kill anyone, and it actually tasted pretty good.
It would certainly suit the initial distribution ranking process. Would it, however, help with the more psychological element of the game - the "A is the best thing I can do but he knows that so he'll expect me to do B but he knows that I'll know that he knows that I know that he knows that we know that I know that he knows ..." (there's a song in there, somewhere)? With games that computers have beaten (or allegedly beaten) the best human players such as Chess and Go, you don't have that element of bluff and counter-bluff. So as much as computational power can only be of help at Colonel Blotto, could Deep Blue ever beat Napoleon?
Whenever you find yourself on the side of the majority, it is time to pause and reflect. - Mark Twain
-
It would certainly suit the initial distribution ranking process. Would it, however, help with the more psychological element of the game - the "A is the best thing I can do but he knows that so he'll expect me to do B but he knows that I'll know that he knows that I know that he knows that we know that I know that he knows ..." (there's a song in there, somewhere)? With games that computers have beaten (or allegedly beaten) the best human players such as Chess and Go, you don't have that element of bluff and counter-bluff. So as much as computational power can only be of help at Colonel Blotto, could Deep Blue ever beat Napoleon?
Whenever you find yourself on the side of the majority, it is time to pause and reflect. - Mark Twain
Yes, it is possible if you don't just train the computer player and then use it forever. Keep a list of them, sorted by their fitness rating and generate a new one from the top 10% for every game. This way the evolutionary process keeps going and will try to adapt to the human player. You should have seen the best of my orcs against human players.
I have lived with several Zen masters - all of them were cats. His last invention was an evil Lasagna. It didn't kill anyone, and it actually tasted pretty good.
-
Slacker007 wrote:
Every human on earth has the propensity for evil
The true question here is what kind of evil. Chaotic evil, neutral evil or lawful evil.
I have lived with several Zen masters - all of them were cats. His last invention was an evil Lasagna. It didn't kill anyone, and it actually tasted pretty good.
Well evil is relative in regards to "type", but the only constant is that evil is the result of negative energy. So does it really matter what kind it is? No, of course not.
-
Yes, it is possible if you don't just train the computer player and then use it forever. Keep a list of them, sorted by their fitness rating and generate a new one from the top 10% for every game. This way the evolutionary process keeps going and will try to adapt to the human player. You should have seen the best of my orcs against human players.
I have lived with several Zen masters - all of them were cats. His last invention was an evil Lasagna. It didn't kill anyone, and it actually tasted pretty good.
A future CP coding challenge, do you think? (albeit with a somewhat simpler game than Blotto - rock, paper, scissors could be a good testing ground).
Whenever you find yourself on the side of the majority, it is time to pause and reflect. - Mark Twain
-
A future CP coding challenge, do you think? (albeit with a somewhat simpler game than Blotto - rock, paper, scissors could be a good testing ground).
Whenever you find yourself on the side of the majority, it is time to pause and reflect. - Mark Twain
Why not? But don't make it too simple. Evolutional algorithms succeed where machine learning fails. You need to know the correct solutions to the problem to train a neural net. What if there is no single correct solution, like in your example here? Or what if you simply don't know the solution, even if it exists? Evolutional algorithms are search algorithms which home in on the best solutions. The drawback is that they also produce very 'weird' solutions, like naturally evolved systems do. There will be one or another appendix which will serve no purpose anymore and sometimes a baby's head will have to go through the mother's pelvis bone at birth.
I have lived with several Zen masters - all of them were cats. His last invention was an evil Lasagna. It didn't kill anyone, and it actually tasted pretty good.
-
I was looking at [this problem](https://puzzling.stackexchange.com/questions/89241/the-100-soldier-problem) which is essentially a game known as [Colonel Blotto](https://en.wikipedia.org/wiki/Blotto\_game) Each general must allocate a given number of troops across a given number of platoons which then battle against each other with results being decided on numerical supremacy. To win you need to both win the greatest number of battles and have the highest number of remaining troops. So what we have is a zero-sum game which has twin objectives (somewhat like hi-lo poker variants but with a requirement to win both hands) with the additional complication of sequencing (a la rock-paper-scissors - e.g. if I set my troops out [19, 1, 19, 1, 19, 1, 19, 1, 19, 1], I might get a very different result than I would be sending out [1, 19, 1, 19, 1, 19, 1, 19, 1, 19] even though I'm using an identical strategy). All of which gets me wondering - how well could we get a machine to play this game against a human opponent? For the first element, calculating the optimal troop split, we could simply brute-force every possible combination, play them against each other and see what wins. Given the vast number of combinations involved, it's obviously a huge task but it's clearly do-able. We're essentially going to wind up with a ranked list of possible distributions). Let's assume that the results of these calculations will be revealed/knowable to the human opponent. Now the human general has an advantage, he knows that the computer knows that the best distribution is distribution X and can easily counter (let's suppose the calculated "optimal" distribution is actually 10 platoons of 10 and the human knows that the computer will play the "best" distribution, the human simply plays 9 platoons of 11 and 1 of 0 and crushes him every time). So the computer needs to throw a few curve balls - the easiest way to do this is to pick a random distribution from the top n distributions. The computer could further obfuscate by applying random increments and decrements to various pairs within the distribution or whatever. Similarly when it comes to ordering the platoons, the computer's most effective approach might just be to randomise the whole distribution array, maybe it can work a more intricate algorithm based on previous distributions employed by the human general. The human of course, has all the same curve balls at his disposal. Machine learning is definitely on the menu and time is no obj
I am doing HPC work with GPUs these days. If it were me, I would use a brute force algorithm written using CUDA to run on a GPU. It's fairly simple algorithm and when running on a few thousand GPU cores it would be pretty fast.
"They have a consciousness, they have a life, they have a soul! Damn you! Let the rabbits wear glasses! Save our brothers! Can I get an amen?"