Electrical Impedance Tomography (EIT) is an inferential method imaging conductivity of domain via current injection and voltage measurement through the finite number of electrodes. Rapid and safe image reconstruction has made EIT a promising imaging method in various fields: medical imaging, robotic skin, flow sensing, and structural health monitoring. However, the modality’s low image quality has significantly hindered practical use. Although people have studied the effect of electrode configuration to improve image quality, however, there have been few systematic approaches to find an optimal electrode placement. This research proposes a novel method to find the optimal configuration of electrodes using active transfer learning and data augmentation. We defined three objective functions related to position, size, and shape distortion of perturbation to evaluate reconstructed images and found optimal electrode placements for each objective function. Furthermore, we propose the optimization framework using transfer learning for the iterative reconstruction method to circumvent tremendous computing processes producing datasets for optimization. The proposed method will be able to optimize electrode configurations under specific environments and objective functions if trainable.