mlsanityData to 14 Oct 2026

B3-Net: Controlled Posterior Bridge Learning for Multi-Task Dense Prediction

arXiv 2605.05722 ↗First posted 7 May 2026cs.CVcited by 0

On the map: State-space sequence models (S4, Mamba) → Vision Mamba backbones

Abstract (arXiv)

Multi-task dense prediction solves complementary pixel-level tasks in a unified model, such as semantic segmentation, depth estimation, surface normal estimation, and edge detection. Existing decoder-side interactions use attention, prompts, routing, diffusion, Mamba, or bridge features to exchange task evidence, but most of them organize this evidence implicitly. They usually fuse task features by similarity or affinity, without explicitly modeling that evidence reliability varies across tasks and spatial locations. As a result, unreliable evidence may contaminate the shared representation and intensify negative transfer. We propose B3-Net, a controlled posterior bridge learning framework for multi-task dense prediction. Our method decomposes decoder-side interaction into reliability estimation, posterior bridge construction, and bounded redistribution. The Precision Field Estimator estimates patch-wise evidence precision from task-reference alignment and local variation. The Posterior Bridge Operator builds a precision-weighted posterior bridge through heteroscedastic evidence fusion, yielding a shared state more reliable than uniform or heuristic mixtures. The Contractive Dispatch Operator redistributes the bridge to each task branch through a bounded update, reducing uncontrolled feature injection. Experiments on NYUD-v2, PASCAL-Context, and Cityscapes show that B3-Net achieves competitive or superior trade-offs over representative CNN-, Transformer-, diffusion-, Mamba-, and bridge-feature-based methods. Backbone-matched comparisons and extensive analyses further verify that the gains arise from controlled posterior bridge learning rather than backbone capacity or decoder scale.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Vision Mamba backbones, one of 10 directions in State-space sequence models (S4, Mamba). Builds on 38 of its 59 references that are in the map. First posted on arXiv 7 May 2026.

Builds on

38 of 59 references.

Closest ancestors · 2

Same direction, newest first

  1. 2024MTMamba++: Enhancing Multi-Task Dense Scene Understanding via Mamba-Based Decoders27 Aug
  2. 2024MTMamba: Enhancing Multi-Task Dense Scene Understanding by Mamba-Based Decoders2 Jul

Other references · 36

  1. 2016The Cityscapes Dataset for Semantic Urban Scene Understandingcited by 4,444
  2. 2017What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?cited by 2,352
  3. 2017Multi-Task Learning Using Uncertainty to Weigh Losses for Scene Geometry and Semanticscited by 1,358
  4. 2020Gradient Surgery for Multi-Task Learningcited by 1,256
  5. 2017GradNorm: Gradient Normalization for Adaptive Loss Balancing in Deep Multitask Networkscited by 834
  6. 2018Multi-Task Learning as Multi-Objective Optimizationcited by 821
  7. 2018Taskonomy: Disentangling Task Transfer Learningcited by 634
  8. 2019Deep Equilibrium Modelscited by 615
  9. 2018End-to-End Multi-Task Learning with Attentioncited by 534
  10. 2016Cross-stitch Networks for Multi-task Learningcited by 467
  11. 2021Conflict-Averse Gradient Descent for Multi-task Learningcited by 459
  12. 2014Detect What You Can: Detecting and Representing Objects using Holistic Models and Body Partscited by 232
24 more
  1. 2018PAD-Net: Multi-Tasks Guided Prediction-and-Distillation Network for Simultaneous Depth Estimation and Scene Parsingcited by 224
  2. 2021Efficiently Identifying Task Groupings for Multi-Task Learningcited by 222
  3. 2022Multi-Task Learning as a Bargaining Gamecited by 215
  4. 2016UberNet: Training a `Universal' Convolutional Neural Network for Low-, Mid-, and High-Level Vision using Diverse Datasets and Limited Memorycited by 200
  5. 2019AdaShare: Learning What To Share For Efficient Deep Multi-Task Learningcited by 160
  6. 2020MTI-Net: Multi-Scale Task Interaction Networks for Multi-Task Learningcited by 157
  7. 2019Attentive Single-Tasking of Multiple Taskscited by 148
  8. 2019Pattern-Affinitive Propagation across Depth, Surface Normal and Semantic Segmentationcited by 142
  9. 2018NDDR-CNN: Layerwise Feature Fusing in Multi-Task CNNs by Neural Discriminative Dimensionality Reductioncited by 127
  10. 2022InvPT: Inverted Pyramid Multi-task Transformer for Dense Scene Understandingcited by 89
  11. 2021Multi-Task Self-Training for Learning General Representationscited by 85
  12. 2022Auto-Lambda: Disentangling Dynamic Task Relationshipscited by 73
  13. 2021Exploring Relational Context for Multi-Task Dense Predictioncited by 71
  14. 2021A Dynamical System Perspective for Lipschitz Neural Networkscited by 60
  15. 2022Factorizing Knowledge in Neural Networkscited by 59
  16. 2024Multi-Task Dense Prediction via Mixture of Low-Rank Expertscited by 58
  17. 2022Multi-Task Learning with Multi-Query Transformer for Dense Predictioncited by 34
  18. 2023InvPT++: Inverted Pyramid Multi-Task Transformer for Visual Scene Understandingcited by 22
  19. 2023Identification of Negative Transfers in Multitask Learning Using Surrogate Modelscited by 20
  20. 2024SwinMTL: A Shared Architecture for Simultaneous Depth Estimation and Semantic Segmentation from Monocular Camera Imagescited by 13
  21. 2023TSP-Transformer: Task-Specific Prompts Boosted Transformer for Holistic Scene Understandingcited by 12
  22. 2023BridgeNet: Comprehensive and Effective Feature Interactions via Bridge Feature for Multi-task Dense Predictionscited by 12
  23. 2020Measuring and Harnessing Transference in Multi-Task Learningcited by 11
  24. 2024Efficient Multitask Dense Predictor via Binarizationcited by 7

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026MUSE: Unlocking Timestep as Native Task Steering for One-Step Dense Prediction
  2. 2025Parameter Aware Mamba Model for Multi-task Dense Prediction
  3. 2024DPBridge: Latent Diffusion Bridge for Dense Prediction
  4. 2025Enhancing Mamba Decoder with Bidirectional Interaction in Multi-Task Dense Prediction
  5. 2025StableMTL: Repurposing Latent Diffusion Models for Multi-Task Learning from Partially Annotated Synthetic Datasets
  6. 2026DeCo: Efficient Decouple-to-Couple Learning for Multi-Task Visual Grounding
  7. 2024DiffusionMTL: Learning Multi-Task Denoising Diffusion Model from Partially Annotated Data
  8. 2024What Matters When Repurposing Diffusion Models for General Dense Perception Tasks?