Neural networks need to start from scratch for each task. They cannot rely on prior knowledge like the human brain can. Google and Alphabet have created a new algorithm for their AI DeepMind, which will allow the network to learn faster. They’re doing it by remembering past events so that it can train on real world data more efficiently.
Just like the human brain, the Elastic Weight Consideration algorithm sets up synapses and stores them when information is useful in relation to previous tasks. The neural networks learn what it needs to make predictions that are more accurate. In a recent set of tests, DeepMind was able to learn the rules from one Atari game and apply that knowledge to the same game.
The tests showed that, of course, adaptation is not easy but the engineers showed that AI could be given memory functions. AI can not only help you with your work, but it can also lead to new discoveries about our own brains.
Learn more about what human neutral network is , by visiting this link : https://aws.amazon.com/what-is/neural-network/.
More information about neutral network , click here .
Author: PC-GR
