5.2 Learning Technologies - Artificial Neural Networks

Learning Technologies - Artificial Neural Networks

So what are artificial neural networks? Well basically its a heuristic technique (1.2 Heuristics) which learners from its previous mistakes, its a more dynamic model in such a way that it makes the AI in a game more intelligent. In sense, your brain is actual a neural network so ANN is just a brain which has been applied to an AI. So how does it work? Simple, software; an artificial network is software which gives the AI its own thinking brain, it's used to think the same as a human player. Whats clever is that the AI will have its own decision tree (4.4 Decision Tree) and learn from its previous experience, for example...

If your playing a 1 v 1 match, player character vs the AI. If you were to stay in the same location\hiding from the AI and manage to kill the AI character in your current location, because the AI character did not know your whereabouts, when the AI re-spawns, from its experience and using artificial neural networks the AI will have learned and will remember where the player character is hiding and wont make the same mistake again by running into your path. This links to (5.3 Learning Technologies - Q-Learning) as the AI apply's what it has learnt from ANN's and made it work in game using Q-Learning.

"the researchers add a phase function to the neural network that essentially prevents it from wrongly mixing different animation types — for instance, taking a step in the middle of a jump."(19)

From Coldeway's comment from his article, it gives us a further look into the artificial neural network mindset, we can start to understand that AI characters are starting to not only think like humans but move and react like them also. We are all for the realism of people and animals in games which uses emergent behaviour (4.3 Boids Flocking) and by adding these featured functions, it allows the developer to insert these different types of animation to make the AI seem as real as possible.

Advantages of ANN's - Well like I mentioned earlier, it allows the AI to learner from its mistake, understand the impact of its role. It understands where its fits in live game play, the AI knows what to do. It uses is environment and current state to determine its next move, this is a clever piece of software which emulates a human down to every feature.

Disadvantage - Well for me, if the AI can think and act as I would then that would only improve the gaming experience and I doubt many would disagree. But obviously the worry could be that by suing ANN's, the AI could learn to out think us and know what we are going to do before we do it.