Transfer Learning with Human Corneal Tissues: An Analysis of Optimal Cut-Off Layer
This addresses a specific challenge in medical image analysis for tissue classification, but it is incremental as it focuses on layer selection within existing architectures.
The study tackled the problem of selecting the optimal cut-off layer in transfer learning for classifying human corneal tissues, finding that middle layers in Inception-v3 and rear layers in VGG-19 provided the best feature representation.
Transfer learning is a powerful tool to adapt trained neural networks to new tasks. Depending on the similarity of the original task to the new task, the selection of the cut-off layer is critical. For medical applications like tissue classification, the last layers of an object classification network might not be optimal. We found that on real data of human corneal tissues the best feature representation can be found in the middle layers of the Inception-v3 and in the rear layers of the VGG-19 architecture.