mlsanityData to 7 Oct 2026

Multi-Task Learning as Multi-Objective Optimization

arXiv 1810.04650 ↗First posted 10 Oct 2018cs.LG stat.MLcited by 809

Not on the map. Why?

Cited by 23 in LLM alignment / RLHF / preference optimisation, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

In multi-task learning, multiple tasks are solved jointly, sharing inductive bias between them. Multi-task learning is inherently a multi-objective problem because different tasks may conflict, necessitating a trade-off. A common compromise is to optimize a proxy objective that minimizes a weighted linear combination of per-task losses. However, this workaround is only valid when the tasks do not compete, which is rarely the case. In this paper, we explicitly cast multi-task learning as multi-objective optimization, with the overall objective of finding a Pareto optimal solution. To this end, we use algorithms developed in the gradient-based multi-objective optimization literature. These algorithms are not directly applicable to large-scale learning problems since they scale poorly with the dimensionality of the gradients and the number of tasks. We therefore propose an upper bound for the multi-objective loss and show that it can be optimized efficiently. We further prove that optimizing this upper bound yields a Pareto optimal solution under realistic assumptions. We apply our method to a variety of multi-task deep learning problems including digit classification, scene understanding (joint semantic segmentation, instance segmentation, and depth estimation), and multi-label classification. Our method produces higher-performing models than recent multi-task learning formulations or per-task training.

Key facts. Cited by 809 papers in the map (data through 7 Oct 2026). First posted on arXiv 10 Oct 2018.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 87; 2022: 97; 2023: 132; 2024: 149; 2025: 171; 2026: 173.

809 papers.

Sort:

By area:

  1. 2021Vision Transformers for Dense Predictioncited by 1,789
  2. 2021Conflict-Averse Gradient Descent for Multi-task Learningcited by 441
  3. 2022Prompt-aligned Gradient for Prompt Tuningcited by 339
  4. 2023Blind Image Quality Assessment via Vision-Language Correspondence: A Multitask Learning Perspectivecited by 276
  5. 2024Simplifying, Stabilizing and Scaling Continuous-Time Consistency Modelscited by 276
  6. 2023AdaMerging: Adaptive Model Merging for Multi-Task Learningcited by 248
  7. 2024Model Merging in LLMs, MLLMs, and Beyond: Methods, Theories, Applications and Opportunitiescited by 247
  8. 2023Efficient Diffusion Training via Min-SNR Weighting Strategycited by 242
  9. 2021MT-Opt: Continuous Multi-Task Robotic Reinforcement Learning at Scalecited by 226
  10. 2021Efficiently Identifying Task Groupings for Multi-Task Learningcited by 218
  11. 2023Sequential Modeling Enables Scalable Learning for Large Vision Modelscited by 212
  12. 2022Multi-Task Learning as a Bargaining Gamecited by 209
38 more
  1. 2021How to Train Your Robot with Deep Reinforcement Learning; Lessons We've Learnedcited by 201
  2. 2023MiDaS v3.1 -- A Model Zoo for Robust Monocular Relative Depth Estimationcited by 173
  3. 2023FengWu: Pushing the Skillful Global Medium-range Weather Forecast beyond 10 Days Leadcited by 170
  4. 2021Motif-based Graph Self-Supervised Learning for Molecular Property Predictioncited by 159
  5. 2022Learning Causally Invariant Representations for Out-of-Distribution Generalization on Graphscited by 144
  6. 2021MSeg: A Composite Dataset for Multi-domain Semantic Segmentationcited by 122
  7. 2024Controllable Preference Optimization: Toward Controllable Multi-Objective Alignmentcited by 105
  8. 2021Reasonable Effectiveness of Random Weighting: A Litmus Test for Multi-Task Learningcited by 100
  9. 2022M3ViT: Mixture-of-Experts Vision Transformer for Efficient Multi-task Learning with Model-Accelerator Co-designcited by 97
  10. 2024Representation Surgery for Multi-Task Model Mergingcited by 95
  11. 2021Reset-Free Reinforcement Learning via Multi-Task Learning: Learning Dexterous Manipulation Behaviors without Human Interventioncited by 90
  12. 2023UniverSeg: Universal Medical Image Segmentationcited by 90
  13. 2022InvPT: Inverted Pyramid Multi-task Transformer for Dense Scene Understandingcited by 89
  14. 2021Multi-Objective Bayesian Optimization over High-Dimensional Search Spacescited by 87
  15. 2021Multi-Objective Loss Balancing for Physics-Informed Deep Learningcited by 86
  16. 2021RotoGrad: Gradient Homogenization in Multitask Learningcited by 82
  17. 2021One Model to Serve All: Star Topology Adaptive Recommender for Multi-Domain CTR Predictioncited by 78
  18. 2024Panacea: Pareto Alignment via Preference Adaptation for LLMscited by 78
  19. 2023FAMO: Fast Adaptive Multitask Optimizationcited by 74
  20. 2022Auto-Lambda: Disentangling Dynamic Task Relationshipscited by 73
  21. 2024MMPareto: Boosting Multimodal Learning with Innocent Unimodal Assistancecited by 72
  22. 2021Exploring Relational Context for Multi-Task Dense Predictioncited by 71
  23. 2021Controlled Text Generation as Continuous Optimization with Multiple Constraintscited by 70
  24. 2021Spinning Language Models: Risks of Propaganda-As-A-Service and Countermeasurescited by 69
  25. 2023Independent Component Alignment for Multi-Task Learningcited by 69
  26. 2022In Defense of the Unitary Scalarization for Deep Multi-Task Learningcited by 65
  27. 2023Advances and Challenges in Meta-Learning: A Technical Reviewcited by 65
  28. 2022Do Current Multi-Task Optimization Methods in Deep Learning Even Help?cited by 60
  29. 2024Distribution Matching for Multi-Task Learning of Classification Tasks: a Large-Scale Study on Faces & Beyondcited by 60
  30. 2021PolyViT: Co-training Vision Transformers on Images, Videos and Audiocited by 58
  31. 2024MetaGPT: Merging Large Language Models Using Model Exclusive Task Arithmeticcited by 53
  32. 2021Pareto Self-Supervised Training for Few-Shot Learningcited by 52
  33. 2022Pareto Set Learning for Expensive Multi-Objective Optimizationcited by 51
  34. 2023Revisiting Scalarization in Multi-Task Learning: A Theoretical Perspectivecited by 51
  35. 2023Recon: Reducing Conflicting Gradients from the Root for Multi-Task Learningcited by 50
  36. 2024Scissorhands: Scrub Data Influence via Connection Sensitivity in Networkscited by 50
  37. 2024Smooth Tchebycheff Scalarization for Multi-Objective Optimizationcited by 50
  38. 2022CodeFill: Multi-token Code Completion by Jointly Learning from Structure and Naming Sequencescited by 49

50 most cited of 809 All 809 (JSON)