Table 1 Characteristics of the Beta-distribution. It is defined with parameters \(\alpha , \beta > 0\) on the interval \([0, 1]\).
From: Stochastic co-teaching for training neural networks with unknown levels of label noise
Parameters | Characteristics |
|---|---|
\(\alpha = \beta\) | Symmetric |
\(\alpha , \beta < 1\) | Bi-modal |
\(\alpha , \beta > 1\) | Uni-modal, bell-shaped |
\(\alpha = \beta = 1\) | Uniform |
\(\alpha < \beta\) | Positively skewed |
\(\alpha > \beta\) | Negatively skewed |
\(\alpha < 1, \beta \ge 1\) | mode at 0 |
\(\alpha = 1, \beta > 1\) | |
\(\alpha \ge 1, \beta < 1\) | mode at 1 |
\(\alpha > 1, \beta = 1\) |
