Sigmoid kernel (SVM)

The sigmoid kernel is a function that can be used in the Kernel trick for SVMs. It has two variables $a, b \in \mathbb{R}_{\geq 0}$. It is defined as

$$K_{a,b}: \mathbb{R}^n \times \mathbb{R}^n \rightarrow \mathbb{R}, \mbox{ by } K_{a,b}(x_1, x_2) = \tanh(a \cdot (x \cdot y) + b).$$