Algorithm 1 Separable Confident Transductive Learning Network. $B(·)$ denotes the mini-batch sets, I is the number of iterations. $p_T={\left\{p_t\right\}}_{t=1}^T$, and T is the number of adjustment learning. $ite{r}_t$ is the t-th adjustment learning.

1:Input: labeled training data ${\mathcal{D}}_{\mathcal{R}}={\left\{({\mathcal{X}}_{\mathcal{R}}^i,{\mathcal{Y}}_{\mathcal{R}}^i)\right\}}_{i=1}^{{\mathcal{N}}_{\mathcal{R}}}$ and unlabeled test data ${\mathcal{D}}_{\mathcal{T}}={\left\{{\mathcal{X}}_{\mathcal{T}}^j\right\}}_{j=1}^{{\mathcal{N}}_{\mathcal{T}}}$ 2:Output: predicted test labels 3:repeat 4:   Derive batch-wise data $(B({\mathcal{X}}_{\mathcal{R}}),B({\mathcal{Y}}_{\mathcal{R}}))$ and $B({\mathcal{X}}_{\mathcal{T}})$ from ${\mathcal{D}}_{\mathcal{R}}$ and ${\mathcal{D}}_{\mathcal{T}}$ 5:   for $iter=1$ to I do 6:        Train classifier f using Equations (1) and (2) 7:        if $iter=ite{r}_t$ then 8:            Get $p_t$ 9:        end if 10:      Generate confident pseudo test labels $\mathcal{C}({\mathcal{Y}}_{{\mathcal{T}}_{\mathcal{P}}^t})$ using Equation (6) 11:      Optimize f using Equation (7) 12:      Minimize the differences between training and test data using Equation (8) 13:      Minimize overall loss with Equation (9) 14:   end for 15:until converged 16:Make prediction for test samples based on trained classifier f