Dynamic Routing Between Capsules

Abstract

A capsule is a group of neurons whose activity vector represents the instantiation parameters of a specific type of entity such as an object or an object part. We use the length of the activity vector to represent the probability that the entity exists and its orientation to represent the instantiation parameters. Active capsules at one level make predictions, via transformation matrices, for the instantiation parameters of higher-level capsules. When multiple predictions agree, a higher level capsule becomes active. We show that a discrimininatively trained, multi-layer capsule system achieves state-of-the-art performance on MNIST and is considerably better than a convolutional net at recognizing highly overlapping digits. To achieve these results we use an iterative routing-by-agreement mechanism: A lower-level capsule prefers to send its output to higher level capsules whose activity vectors have a big scalar product with the prediction coming from the lower-level capsule.

References in corpus (7)

• Adam: A Method for Stochastic Optimization
• TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems
• Multiple Object Recognition with Visual Attention
• Stochastic Pooling for Regularization of Deep Convolutional Neural Networks
• High-Performance Neural Networks for Visual Object Classification
• APAC: Augmented PAttern Classification with Neural Networks
• Batch-normalized Maxout Network in Network

Cited by in corpus (34)

• Explainable artificial intelligence (XAI) in deep learning-based medical image analysis
• Capsule Network Performance on Complex Data
• An overview of artificial intelligence techniques for diagnosis of Schizophrenia based on magnetic resonance imaging modalities: Methods, challenges, and future works
• An Introduction to Image Synthesis with Generative Adversarial Nets
• Distance-based Self-Attention Network for Natural Language Inference
• Fanaroff-Riley classification of radio galaxies using group-equivariant convolutional neural networks
• Graph Capsule Aggregation for Unaligned Multimodal Sequences
• Centroid Transformers: Learning to Abstract with Attention
• Self-Attention Capsule Networks for Object Classification
• Towards Generalizable Surgical Activity Recognition Using Spatial Temporal Graph Convolutional Networks
• Speech recognition with quaternion neural networks
• Autonomous Deep Learning: A Genetic DCNN Designer for Image Classification
• E2-Capsule Neural Networks for Facial Expression Recognition Using AU-Aware Attention
• ToyArchitecture: Unsupervised Learning of Interpretable Models of the World
• Topographic VAEs learn Equivariant Capsules
• Multi-modal anticipation of stochastic trajectories in a dynamic environment with Conditional Variational Autoencoders
• AOGNets: Compositional Grammatical Architectures for Deep Learning
• Transformer visualization via dictionary learning: contextualized embedding as a linear superposition of transformer factors
• In folly ripe. In reason rotten. Putting machine theology to rest
• Deep Sparse Coding for Invariant Multimodal Halle Berry Neurons
• Towards Adversarial Training with Moderate Performance Improvement for Neural Network Classification
• How do SGD hyperparameters in natural training affect adversarial robustness?
• Development of Conditional Random Field Insert for UNet-based Zonal Prostate Segmentation on T2-Weighted MRI
• Spiking CapsNet: A Spiking Neural Network With A Biologically Plausible Routing Rule Between Capsules
• Bio-inspired Robustness: A Review
• Three-Dimensional Fourier Scattering Transform and Classification of Hyperspectral Images
• Interpreting A Pre-trained Model Is A Key For Model Architecture Optimization: A Case Study On Wav2Vec 2.0
• Hater-O-Genius Aggression Classification using Capsule Networks
• Vision System for AGI: Problems and Directions
• Deep Learning: a new definition of artificial neuron with double weight
• A New Framework for Machine Intelligence: Concepts and Prototype
• Recurrent Attention Models with Object-centric Capsule Representation for Multi-object Recognition
• Security Analysis of Capsule Network Inference using Horizontal Collaboration
• Using Artificial Intelligence to Augment Science Prioritization for Astro2020