2019/07/29 by Steven J. Frank, Frank, Steven J., Andrea Frank +1 · 1 citation
Computer Science · #Advanced Neural Network Applications #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #FOS: Electrical engineering #Generative Adversarial Networks and Image Synthesis #Image and Video Processing (eess.IV) #Machine Learning (cs.LG) #Medical Image Segmentation Techniques #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.1907.12436
openalex publication_date 2019/07/29 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
As a training and analysis strategy for convolutional neural networks (CNNs),\nwe slice images into tiled segments and use, for training and prediction,\nsegments that both satisfy a criterion of information diversity and contain\nsufficient content to support classification. In particular, we utilize image\nentropy as the diversity criterion. This ensures that each tile carries as much\ninformation diversity as the original image, and for many applications serves\nas an indicator of usefulness in classification. To make predictions, a\nprobability aggregation framework is applied to probabilities assigned by the\nCNN to the input image tiles. This technique facilitates the use of large,\nhigh-resolution images that would be impractical to analyze unmodified;\nprovides data augmentation for training, which is particularly valuable when\nimage availability is limited; and the ensemble nature of the input for\nprediction enhances its accuracy.\n