sources of bias

Terms from Artificial Intelligence: humans at the heart of algorithms

Page numbers are for draft copy at present; they will be replaced with correct numbers when final book is formatted. Chapter numbers are correct and will not change now.

It is known that many machine learning systems can have bias. There are various sources for this. One of the most obvious is human bias captured in the training data. In addition, the choice of {[fitness function}} and input features both have an impact, as do base rate differences, often themselves due to pre-existing societal bias.

Used on Chap. 20: page 494

Bias entering at different stages in machine learning.