Incorporating the Knowledge Distillation to Improve the EfficientNet Transfer Learning Capability
Due to the Deep Learning requirement for a large training dataset, Transfer Learning has become a central method in the field of Computer Vision, which heavily used Deep Learning. Since the adoption of transfer learning in the field, the performance of the models in computer vision is significantly upgraded. In this paper, we proposed a transfer learning method that can further improve the accuracy of the EfficientNet, the current state-of-the-art image classification model. The proposed method is able to upgrade the performance of each of the EfficientNet architecture that can even outperform the larger architecture that is trained using the standard transfer learning method.
Tjeng Wawan Cenggoro
