Personally, I think they're better then using a pure state based approach for strategy games' fight or flight decisions, but it really depends on the complexity.. Other then that, I can't think of somewhere where it's easiest, or worthwhile at all to implement something like this.
Maybe after it's been explored for a while and there's a multitude of third party API's to work with them easier, assuming it's worth creating those (Which I doubt it is).
so you dont like exploring new ideas, youd rather everyone else do that for you.
to each their own i guess.
so you dont like exploring new ideas, youd rather everyone else do that for you.
to each their own i guess.
And with this, you have jumped the proverbial shark.
ANNs are not even remotely "new". They are also hopelessly over-hyped. They are a simple pattern-matching algorithm poorly equipped to handle most of the situational reasoning tasks necessary for game AI or many other kinds of AI applications. Even in the traditional machine learning arena, NNs are somewhat considered passe since there are far better alternatives.
But I see you don't like researching ideas beyond what you Obviously Know Is Best. You'd rather everyone else do that for you.
To each his, own I guess.
Personally, I think they're better then using a pure state based approach for strategy games' fight or flight decisions, but it really depends on the complexity.. Other then that, I can't think of somewhere where it's easiest, or worthwhile at all to implement something like this.
Actually, while I agree that pure state based approaches in strategy games are a problem, a NN isn't the answer really. When dealing with a huge influx of variable data such as we have in more complex situations, a utility-based approach is by far preferable. As with a NN, there are a ton of numbers being balanced and played against one another. The difference is, in a utiltiy-based system, you know what those numbers are doing with their relationships to each other. An NN, by its very nature, is opaque. You can't even lift the lid and point to something and say, "and this is why it made that decision". There is no such entity to point at in an NN. It just comes out the meat grinder.
They're both number crunchers... one method is based on intelligent design -- the other on hope and fairies.
What if the inputs of the neural network were not just an enemy character in the game, but inputs for how the screen scrolls and zooms, enemies are spawned, enemies react, die, what happens to the player when he touches different things... basicly every single activity that happens during the game.
I was thinking you could train it all with backpropagation, you actually hand feed the neural network the motions of the game as if it was playing, then after it is trained it should be able to run the game for you.
I don't want to add to the previous heartfelt advice against neural network misapplication, because stubbornly defying contrary opinions might still be a useful learning experience; this description of what you want to do is much more worrying than the risk of wasting time by trying an inadequate technique, because you don't want to try a technique, you want a magic wand. Wishful thinking and learning rarely mix.
While trying to develop a bot to play your game is a fine objective, you don't seem to approach it on a sound problem-solving basis, hoping instead that applying a neural network is going to be easy and effective: this sort of a priori preference for a certain solution is the opposite of good engineering and design, and it would be equally bad in the case of a good technique.
You don't even state clearly what sort of game you are thinking of, neglecting the analysis of what an AI for your game needs to be able to do and what are the difficulties and non-difficulties in such tasks, which is the first step in choosing appropriate AI architectures and algorithms and/or modifying game rules to make the AI perform better (for example, simplifying game state to reduce the amount of training needed and facilitate unsupervised trial and error learning). Do you expect this work to disappear?
