This handout illustration obtained from DeepMind on May 30, 2019 shows agents playing Capture the Flag, showing a range of behaviors. "AFP PHOTO / DEEPMIND "
Your feedback is important to us!
We invite all our readers to share with us their views and comments about this article.
Disclaimer: Comments submitted by third parties on this site are the sole responsibility of the individual(s) whose content is submitted. The Daily Star accepts no responsibility for the content of comment(s), including, without limitation, any error, omission or inaccuracy therein. Please note that your email address will NOT appear on the site.
Alert: If you are facing problems with posting comments, please note that you must verify your email with Disqus prior to posting a comment. follow this link to make sure your account meets the requirements. (http://bit.ly/vDisqus)
A team of programmers at a British artificial intelligence company has designed automated "agents" that taught themselves how to play the seminal first-person shooter Quake III Arena and became so good they consistently beat human beings.More recently, a Google AI agent beat the world's No. 1 Go player in 2017 ."One of the contributions of the paper is each agent learns its own internal reward signal," Jaderbeg said, meaning that the AI players decided for themselves how much weight to assign the successful completion of tasks like capturing the flag or hitting an opponent.Next, they found that training a population of agents together, rather than one at a time, made the population as a whole learn much faster.
FOLLOW THIS ARTICLE