Boosted Training of Convolutional Neural Networks for Multi-Class Segmentation
This work addresses computational and memory efficiency problems for researchers and practitioners in medical imaging segmentation, though it appears incremental as it builds on existing methods with specific enhancements.
The paper tackles the challenge of training deep neural networks on large, sparse datasets for semantic segmentation, such as CT scans, by proposing a boosted sampling scheme and adaptive learning rate algorithm, resulting in significant training speed-up and new state-of-the-art results on the VISCERAL Anatomy benchmark.
Training deep neural networks on large and sparse datasets is still challenging and can require large amounts of computation and memory. In this work, we address the task of performing semantic segmentation on large volumetric data sets, such as CT scans. Our contribution is threefold: 1) We propose a boosted sampling scheme that uses a-posterior error maps, generated throughout training, to focus sampling on difficult regions, resulting in a more informative loss. This results in a significant training speed up and improves learning performance for image segmentation. 2) We propose a novel algorithm for boosting the SGD learning rate schedule by adaptively increasing and lowering the learning rate, avoiding the need for extensive hyperparameter tuning. 3) We show that our method is able to attain new state-of-the-art results on the VISCERAL Anatomy benchmark.