Algorithm 1 GP-MI.

${\widehat{\gamma }}_0\leftarrow 0$ for $t=1,2,\dots$ do  Compute ${\mu }_t$ and ${\sigma }_t^2$ by ${\chi }_t=\left\{x_1..x_{t-1}\right\}$ and ${\mathcal{Y}}_t=\left(y_1..y_{t-1}\right)$ // Bayesian inference  ${\varphi }_t\left(x\right)\leftarrow \sqrt{log{\displaystyle \frac{2}{\delta }}}\left(\sqrt{{\sigma }_t^2\left(x\right)+{\widehat{\gamma }}_{t-1}}-\sqrt{{\widehat{\gamma }}_{t-1}}\right)$ // Definition of ${\varphi }_t\left(x\right)$ for all $x\in \chi$  $x_t\leftarrow \underset{x\in \chi }{argmax}{\mu }_t\left(x\right)+{\varphi }_t\left(x\right)$ // Selection of the next query location  ${\widehat{\gamma }}_t\leftarrow {\widehat{\gamma }}_{t-1}+{\sigma }_t^2\left(x_t\right)$ // Update ${\widehat{\gamma }}_t$  get $y_t$ by the DNILMF model and $x_t$ // Query $\left(x_t,y_t\right)$ end for