Hi!
I'm thinking of developing some kind of a strategy game which will learn how the player plays as the time goes by.
It will probably have to consider a lot of variables in order to reach his best move each time and win.

But i'm a little confused about which learning method I should choose...
which would you recommend?
thanks! :)

Players like to win most of the time. I don't recommend trying to beat the player, but instead, trying to challenge the player in a way that maximizes enjoyment from the game, which means playing in such a way that the player still usually wins, but there is at least a consistent and evolving failure state that does happen sometimes and will happen if the player doesn't stay on his/her toes.

As to the strategy game: It depends entirely on the rules of the game itself. There is no one-size-fits-all plug and play AI solution. You kind of have to design the game to be easy for the computer to analyze and solve.

Can you tell us more about your game design?

what do you mean by "game design"?

what do you mean by "game design"?

Details about the features you intend to implement.

It's supposed to be a normal strategy game with some extra features...
for example the player would be able to build a bridge in order to mobilize its units to areas it couldn't access without it.
I'd like the AI to know that unless the user shows signs of building that bridge, it won't build up it defences in that area or if the user shows interest in some region, the AI will be ready for him.

Design an AI that plays the game first. Until you do that, there is no "learning" that can happen.

Ok, but I'm doing some research now so I'll know what I can and can't do. :)