Which algorithm for my AI ?
Which algorithm for my AI ?
Hi,
I'm Jérôme, a student in IT from France. I'm currently on a risk-like project for my school. We're actually 4 people working on this project, and I'm in charge of the AI part of our game.
Well, I was wondering on the internet when I found your game. It is very interesting and I thought that you might be able to give some advices about which algorithm could be the best for our game.
First I thought about a neural network, and now I'm trying a min-max algorithm. However it seems the the min-max algorithm is more dedicated to 2-players game.
Could you help me on that point. And if any general advice about how to make an AI, feel free to add them to your reply. You may have to know that we use a static map in our game.
Thanks
I'm Jérôme, a student in IT from France. I'm currently on a risk-like project for my school. We're actually 4 people working on this project, and I'm in charge of the AI part of our game.
Well, I was wondering on the internet when I found your game. It is very interesting and I thought that you might be able to give some advices about which algorithm could be the best for our game.
First I thought about a neural network, and now I'm trying a min-max algorithm. However it seems the the min-max algorithm is more dedicated to 2-players game.
Could you help me on that point. And if any general advice about how to make an AI, feel free to add them to your reply. You may have to know that we use a static map in our game.
Thanks
Bonjour Jérôme,
My advice is to forget fancy algorithms, and go with simple common sense. Risk is not "deterministic" like chess or checkers because of the dice. I think that min-max can not work because there is no way to predict what will happen.
The following rule applies in AI design: it always takes more time than you think. Much longer. Use this technique to figure out the time you will need to design an AI:
- Estimate the time required to code and debug your AI, by listing everything you can think of;
- Then, multiply that by 2;
- Then, add 25%;
- Since you are not working from a stable platform, multiply by 2 again.
This may seem like a joke, but trust me. It took me between 500 and 1000 hours to design Reaper. Contrast this with the time you have available. Conclusion: go for something simple, like the communist bot in lux. With that as a goal, you have a good chance of having a functionning AI at the end of your project. You can always refine it later if you have some time left.
Bonne Chance...
My advice is to forget fancy algorithms, and go with simple common sense. Risk is not "deterministic" like chess or checkers because of the dice. I think that min-max can not work because there is no way to predict what will happen.
The following rule applies in AI design: it always takes more time than you think. Much longer. Use this technique to figure out the time you will need to design an AI:
- Estimate the time required to code and debug your AI, by listing everything you can think of;
- Then, multiply that by 2;
- Then, add 25%;
- Since you are not working from a stable platform, multiply by 2 again.
This may seem like a joke, but trust me. It took me between 500 and 1000 hours to design Reaper. Contrast this with the time you have available. Conclusion: go for something simple, like the communist bot in lux. With that as a goal, you have a good chance of having a functionning AI at the end of your project. You can always refine it later if you have some time left.
Bonne Chance...
Who is online
Users browsing this forum: Google [Bot] and 58 guests