Learn more Download now Shop now Shop now Shop now Shop now Learn More Shop now Shop now Learn more Shop Fire Shop Kindle Learn More Shop now Shop now Learn more

Customer reviews

5.0 out of 5 stars
2

on 26 November 2014
I adore this book - I really enjoyed of programming wisdom (the previous lineup from Rabin) and I really enjoyed this, but that enjoyment is because I got what I expected.

I shall demonstrate this with an example.

I did not know that the book contained information on using Neural Networks to evaluate the utility of different actions for agents in an RTS. However I now know that in Supreme Commander II a Neural Network was used to evaluate whether or not the AI should retreat, gather resources, attack the weakest, attack the strongest ... so forth (Page 391) I now know that programmer's experience in doing this, some of the pit falls, I also know that there is at least 1 successful implementation of NN in games and how well it worked.

In SCII a victory condition is to destroy the Armoured Command Unit (ACU - on the front cover) - doing so causes a huge nuclear explosion that damages everything near by, so a fun bit of trivia I learnt: initially the AI would run away from the ACU and not attack, because it would suffer losses, I now know how the fitting curve was SUCCESSFULLY adjusted so the AI recognises the victory condition.

I had no idea I'd find that out going in, but I do have a working knowledge (and could implement with minor book-referencing) of a few types of neural networks.

I also know thanks to Rabin's earlier books (Wisdom, book 1 I think) that neural networks were used (unsuccessfully - mostly) in Black and White, for the Creature AI - I say unsuccessfully because in a patch to the game a tool-tip was added to make training the creature easier.

That is a great example because the paper tells me about how it actually happened in Supreme Commander II, I don't need to know what a neural network is it's a story essentially.

So this book is for those who already know the internals of game engines and are comfortable with combinatorial optimisation problems and dare I say being an experienced programmer helps, these made the book VERY enjoyable for me and have warned/demonstrated things that I may build upon myself.

I really recommend this book, without hesitation it is a good book, make sure you are in the intended audience before you buy though.

If you are, you simply must get this book, it's fascinating, it's like news and a story of something interesting combined.

I hope this helps someone, please leave me a comment if you have any feedback, I'd like to get better at writing reviews.
0Comment| 7 people found this helpful. Was this review helpful to you? Report abuse
on 12 May 2014
one of the best AI books that simply states the facts and where and when to use these fact. The chapters on Hierarchical Task Network's are great and fairly easy to understand.

Great book and worth the £50.
0Comment| One person found this helpful. Was this review helpful to you? Report abuse



Need customer service? Click here

Sponsored Links

  (What is this?)