Mode-locking Theory for Long-Range Interaction in Artificial Neural Networks
Not on the map. Why?
Abstract (arXiv)
Visual long-range interaction refers to modeling dependencies between distant feature points or blocks within an image, which can significantly enhance the model's robustness. Both CNN and Transformer can establish long-range interactions through layering and patch calculations. However, the underlying mechanism of long-range interaction in visual space remains unclear. We propose the mode-locking theory as the underlying mechanism, which constrains the phase and wavelength relationship between waves to achieve mode-locked interference waveform. We verify this theory through simulation experiments and demonstrate the mode-locking pattern in real-world scene models. Our proposed theory of long-range interaction provides a comprehensive understanding of the mechanism behind this phenomenon in artificial neural networks. This theory can inspire the integration of the mode-locking pattern into models to enhance their robustness.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 33 of its 54 references that are in the map. First posted on arXiv 10 Mar 2023.
Builds on
33 of 54 references.
References · 33
- 2015Deep Residual Learning for Image Recognitioncited by 51,809
- 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
- 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,267
- 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,082
- 2016Semi-Supervised Classification with Graph Convolutional Networkscited by 11,282
- 2017Graph Attention Networkscited by 8,854
- 2016Densely Connected Convolutional Networkscited by 6,691
- 2017Inductive Representation Learning on Large Graphscited by 6,284
- 2014Fully Convolutional Networks for Semantic Segmentationcited by 4,845
- 2018MobileNetV2: Inverted Residuals and Linear Bottleneckscited by 4,773
- 2014Going Deeper with Convolutionscited by 4,231
- 2016DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFscited by 3,239
21 more
- 2018Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentationcited by 3,210
- 2017Non-local Neural Networkscited by 2,290
- 2021Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutionscited by 1,960
- 2019MMDetection: Open MMLab Detection Toolbox and Benchmarkcited by 1,379
- 2016Variational Graph Auto-Encoderscited by 1,373
- 2017DeepAR: Probabilistic Forecasting with Autoregressive Recurrent Networkscited by 1,004
- 2014Deep Visual-Semantic Alignments for Generating Image Descriptionscited by 966
- 2021CrossViT: Cross-Attention Multi-Scale Vision Transformer for Image Classificationcited by 806
- 2019Panoptic Feature Pyramid Networkscited by 644
- 2015Holistically-Nested Edge Detectioncited by 590
- 2017A Multi-Horizon Quantile Recurrent Forecastercited by 211
- 2019Dice Loss for Data-imbalanced NLP Taskscited by 180
- 2016Richer Convolutional Features for Edge Detectioncited by 141
- 2014Fast Edge Detection Using Structured Forestscited by 64
- 2016DeepSkeleton: Learning Multi-task Scale-associated Deep Side Outputs for Object Skeleton Extraction in Natural Imagescited by 13
- 2017SRN: Side-output Residual Network for Object Symmetry Detection in the Wildcited by 10
- 2016Object Skeleton Extraction in Natural Images by Fusing Scale-associated Deep Side Outputscited by 9
- 2020Adaptive Linear Span Network for Object Skeleton Detectioncited by 9
- 2018Linear Span Network for Object Skeleton Detectioncited by 7
- 2018Hi-Fi: Hierarchical Feature Integration for Skeleton Detectioncited by 6
- 2022Emergence of Double-slit Interference by Representing Visual Space in Artificial Neural Networkscited by 1
Built on by
Not cited yet.