mlsanityData to 7 Oct 2026

Deciding How to Decide: Dynamic Routing in Artificial Neural Networks

arXiv 1703.06217 ↗First posted 17 Mar 2017stat.ML cs.CV cs.LGcited by 33

Not on the map. Why?

Abstract (arXiv)

We propose and systematically evaluate three strategies for training dynamically-routed artificial neural networks: graphs of learned transformations through which different input signals may take different paths. Though some approaches have advantages over others, the resulting networks are often qualitatively similar. We find that, in dynamically-routed networks trained to classify images, layers and branches become specialized to process distinct categories of images. Additionally, given a fixed computational budget, dynamically-routed networks tend to perform better than comparable statically-routed networks.

Key facts. Cited by 33 papers in the map (data through 7 Oct 2026). First posted on arXiv 17 Mar 2017.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 10; 2022: 7; 2023: 6; 2024: 4; 2025: 3; 2026: 3.

33 papers.

Sort:

By area:

  1. 2021Dynamic Neural Networks: A Surveycited by 405
  2. 2021AdaFuse: Adaptive Temporal Fusion Network for Efficient Action Recognitioncited by 48
  3. 2021Zero Time Waste: Recycling Predictions in Early Exit Neural Networkscited by 39
  4. 2021AdaMML: Adaptive Multi-Modal Learning for Efficient Video Recognitioncited by 36
  5. 2021Dynamic Network Quantization for Efficient Video Inferencecited by 36
  6. 2025VORTA: Efficient Video Diffusion via Routing Sparse Attentioncited by 23
  7. 2021Anytime Dense Prediction with Confidence Adaptivitycited by 18
  8. 2024TESTAM: A Time-Enhanced Spatio-Temporal Attention Model with Mixture of Expertscited by 18
  9. 2022Neural Networks are Decision Treescited by 16
  10. 2021VA-RED2: Video Adaptive Redundancy Reductioncited by 15
  11. 2023DynamicDet: A Unified Dynamic Architecture for Object Detectioncited by 15
  12. 2022Spatial Mixture-of-Expertscited by 14
21 more
  1. 2023Revisiting Single-gated Mixtures of Expertscited by 10
  2. 2021E2CM: Early Exit via Class Means for Efficient Supervised and Unsupervised Learningcited by 8
  3. 2022Boosted Dynamic Neural Networkscited by 8
  4. 2023AsConvSR: Fast and Lightweight Super-Resolution Network with Assembled Convolutionscited by 8
  5. 2022A Survey of Neural Treescited by 7
  6. 2022Understanding the Robustness of Multi-Exit Models under Common Corruptionscited by 4
  7. 2025EgoAdapt: Adaptive Multisensory Distillation and Policy Learning for Efficient Egocentric Perceptioncited by 4
  8. 2026ODAR: Principled Adaptive Routing for LLM Reasoning via Active Inferencecited by 4
  9. 2021Sparsely ensembled convolutional neural network classifiers via reinforcement learningcited by 2
  10. 2021Complexity-aware Adaptive Training and Inference for Edge-Cloud Distributed AI Systemscited by 2
  11. 2023ATESA-BAERT: A Heterogeneous Ensemble Learning Model for Aspect-Based Sentiment Analysiscited by 2
  12. 2025Divide, Specialize, and Route: A New Approach to Efficient Ensemble Learningcited by 2
  13. 2022Transition Matrix Representation of Trees with Transposed Convolutionscited by 1
  14. 2022Neural Routing in Meta Learningcited by 0
  15. 2023Unifying Synergies between Self-supervised Learning and Dynamic Computationcited by 0
  16. 2023LEURN: Learning Explainable Univariate Rules with Neural Networkscited by 0
  17. 2024Conditional Information Gain Trelliscited by 0
  18. 2024Rule Based Learning with Dynamic (Graph) Neural Networkscited by 0
  19. 2024Block-Operations: Using Modular Routing to Improve Compositional Generalizationcited by 0
  20. 2026cMoLLM at Scale: Horizontal Scaling Laws for Mixture-of-LLMscited by 0
  21. 2026Harnessing Domain Specialists in Multimodal Mixture-of-Experts for Efficient Adaptationcited by 0