 |
This article is orphaned as few or no other articles link to it. Please help introduce links in articles on related topics. (September 2006) |
Alpha (α) is a parameter used in artificial intelligence to specify how much weight will be given to new information. An alpha value might be expressed in the range between 0 and 1. A value of 0 means that the agent completely ignores any new information, while a value of 1 means it only remembers the latest piece of information. A typical equation detailing the weighting role of alpha might be:
new_policy = (historical_data * (1 - α)) + (new_data * α)
It is used in Q-learning.
