Algorithm 2 Pseudo Code for ODAELM-T.

Input:

$L:=$ the number of hidden layer neurons;

$Act:=$ the activation function type;

$SD:=$ the source domain data;

1: Initialize labeled and unlabeled set as $LSet$ and $USet$, respectively.

2: Initialize the source classifier of L hidden nodes using $ActType$ with $SD$;

3: Let $H_S$ and ${\beta }_S$ be defined as Equation (17);

4: while new sample x in the target domain arrives do

5: Calculate the probability P for labeling;

6: Generate random value between 0 and 1 as p;

7: if $p<P$ then

8:  Add x into $USet$;

9:  Select a group of samples as $\delta Set$ in the target domain for labeling;

10:  if $LSet$ is empty then

11:   $LSet=\delta Set$;

12:   Initialize a target classifier of L hidden nodes using Equation (17);

13:  else

14:   perform unlabeled incremental learning where increment is x;

15:   $USet=USet-\delta Set$;

16:    perform unlabeled decremental learning where decrement is $\delta Set$;

17:   when the labeling process completes, $LSet=LSet+\delta Set$;

18:   perform labeled incremental learning where increment is $\delta Set$;

19:  end if

20: else

21:  Add x into $USet$;

22:  perform unlabeled incremental learning where the increment is x;

23: end if

24: end while