Deviate from theory.

Bias is a common English word, which is involved in many places of signal detection and estimation. It is difficult to give a detailed explanation without context. Detection problems are generally described by percentages such as detection rate and false negative rate, while estimation problems are generally described by mean square error (MSE) and cross-correlation coefficient (R2), and there is no commonly used description method called bias. The possible place is the nature of estimation or the boundary of clem. When defining unbiased estimation, bias is specifically defined as deviation, and then bias is defined as the difference between the function mean of estimator and the sample mean. When the deviation is always zero, it is unbiased estimation, when the sample tends to infinity, it is progressive unbiased estimation, otherwise it is biased estimation. Other places involving deviation, such as minimax detection criterion and least square estimation criterion, may also be encountered. Find the place where the word appears, find the context, and see how the things described are defined, so you should know how prejudice is defined.