Skip to main content

Currently Skimming:

Creating Intelligent Agents in Games
Pages 15-28

The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.


From page 15...
... Inexpensive yet powerful computer hardware has made it possible to simulate complex physical environments, resulting in tremendous growth in the video game industry. From modest sales in the 1960s (Baer, 2005)
From page 16...
... Therefore, just as symbolic games provided an opportunity for the development and testing of GOFAI techniques in the 1980s and 1990s, video games provide an opportunity for the development and testing of machine-learning techniques and their transfer to industry. ARTIFICIAL INTELLIGENCE IN VIDEO GAMES One of the main challenges for AI is creating intelligent agents that can become more proficient in their tasks over time and adapt to new situations as they occur.
From page 17...
... This technique not only promises to rise to the challenge of creating games that are educational, but also promises to provide a platform for the safe, effective study of how intelligent agents adapt. Evolutionary computation is a computational machine-learning technique modeled after natural evolution (Figure 1a)
From page 18...
... Solutions (such as neural networks) are encoded as chromosomes, usually consisting of strings of real numbers, in a population.
From page 19...
... , which was originally developed for learning behavioral strategies. The neural networks control agents that select actions in their output based on sensory inputs (Figure 1b)
From page 20...
... This cycle of removal and replacement happens continually throughout the game and is largely invisible to the player. As a result, the algorithm can evolve increasingly complex neural networks fast enough for a user to interact with evolution as it happens in real time.
From page 21...
... prepare them for combat, the player must design a sequence of training exercises and goals. Ideally, the exercises will be increasingly difficult so that the team begins by learning basic skills and then gradually builds on them (Figure 2)
From page 22...
... Each agent is controlled by a neural network with random connection weights and no hidden nodes, which is the usual starting configuration for NEAT. As the neural networks are replaced in real time, behavior improves, and agents eventually learn to perform the task the player has set up.
From page 23...
... b. Incremental training on increasingly complex wall configurations produced agents that could navigate this complex maze to find the enemy.
From page 24...
... For example, a seeking team won six out of ten battles, only a slight advantage, against an avoidant team that ran in a pack to a corner of the field next to an enclosing wall. Sometimes, if an avoidant team made it to the corner and assembled fast enough, the seeking team ran into an ambush and was obliterated.
From page 25...
... Just as traditional symbolic games catalyzed the development of GOFAI techniques, video gaming may catalyze research in machine learning for decades to come. ACKNOWLEDGMENTS This work was supported in part by the Digital Media Collaboratory of the University of Texas at Austin, Texas Higher Education Coordinating Board, through grant ARP-003658-476-2001, and the National Science Foundation through grants EIA-0303609 and IIS-0083776.
From page 26...
... 2002. Evolving neural networks through augmenting topologies.
From page 27...
... 1999. Evolving artificial neural networks.


This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More information on Chapter Skim is available.