In-Model Merging for Enhancing the Robustness of Medical Imaging Classification Models
This work addresses the critical need for robust models in medical image analysis, offering an incremental advancement over existing model merging methods by focusing on internal kernel merging.
The paper tackles the problem of enhancing robustness in medical imaging classification models by proposing in-model merging (InMerge), a technique that selectively merges similar convolutional kernels within a single CNN during training, resulting in substantial performance improvements over typically-trained models on four datasets.
Model merging is an effective strategy to merge multiple models for enhancing model performances, and more efficient than ensemble learning as it will not introduce extra computation into inference. However, limited research explores if the merging process can occur within one model and enhance the model's robustness, which is particularly critical in the medical image domain. In the paper, we are the first to propose in-model merging (InMerge), a novel approach that enhances the model's robustness by selectively merging similar convolutional kernels in the deep layers of a single convolutional neural network (CNN) during the training process for classification. We also analytically reveal important characteristics that affect how in-model merging should be performed, serving as an insightful reference for the community. We demonstrate the feasibility and effectiveness of this technique for different CNN architectures on 4 prevalent datasets. The proposed InMerge-trained model surpasses the typically-trained model by a substantial margin. The code will be made public.