Robust neural circuit reconstruction from serial electron microscopy with convolutional recurrent networks
arXiv:1811.11356
Abstract
Recent successes in deep learning have started to impact neuroscience. Of particular significance are claims that current segmentation algorithms achieve "super-human" accuracy in an area known as connectomics. However, as we will show, these algorithms do not effectively generalize beyond the particular source and brain tissues used for training -- severely limiting their usability by the broader neuroscience community. To fill this gap, we describe a novel connectomics challenge for source- and tissue-agnostic reconstruction of neurons (STAR), which favors broad generalization over fitting specific datasets. We first demonstrate that current state-of-the-art approaches to neuron segmentation perform poorly on the challenge. We further describe a novel convolutional recurrent neural network module that combines short-range horizontal connections within a processing stage and long-range top-down connections between stages. The resulting architecture establishes the state of the art on the STAR challenge and represents a significant step towards widely usable and fully-automated connectomics analysis.
References in corpus (23)
- Adam: A Method for Stochastic Optimization
- Understanding deep learning requires rethinking generalization
- Recurrent Residual Convolutional Neural Network based on U-Net (R2U-Net) for Medical Image Segmentation
- Deep Predictive Coding Networks for Video Prediction and Unsupervised Learning
- Alleviating catastrophic forgetting using context-dependent gating and synaptic stabilization
- Recurrent computations for visual pattern completion
- Superhuman Accuracy on the SNEMI3D Connectomics Challenge
- Combining Fully Convolutional and Recurrent Neural Networks for 3D Biomedical Image Segmentation
- Measuring the tendency of CNNs to Learn Surface Statistical Regularities
- Bridging the Gaps Between Residual Learning, Recurrent Neural Networks and Visual Cortex
- Machine learning of hierarchical clustering to segment 2D and 3D images
- Why do deep convolutional networks generalize so poorly to small image transformations?
- On orthogonality and learning recurrent networks with long term dependencies
- Comparing deep neural networks against humans: object recognition when the signal gets weaker
- Task-Driven Convolutional Recurrent Models of the Visual System
- Improving Deep Pancreas Segmentation in CT and MRI Images via Recurrent Neural Contextual Learning and Direct Loss Function
- Do CIFAR-10 Classifiers Generalize to CIFAR-10?
- Deep Predictive Coding Network for Object Recognition
- Bi-Directional Cascade Network for Perceptual Edge Detection
- Deep Predictive Learning: A Comprehensive Model of Three Visual Streams
- Can recurrent neural networks warp time?
- Visual Concepts and Compositional Voting
- Guided Proofreading of Automatic Segmentations for Connectomics