Softmax

machine-learning

Definition

Softmax

The softmax function $\sigma$ is a function that takes a vector $\mathbf{x}\in {\mathbb{R}}^n$ and maps it into a probability distribution of $n$ possible outcomes:

$$\sigma (\mathbf{x})=\frac{e^{\mathbf{x}}}{{\sum }_{i=1}^n{e}^{{\mathbf{x}}_i}}$$