Research map / Learning theory and algorithms / Combinatorial and search optimization
Efficient Continuous Pareto Exploration in Multi-Task Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Apply Pareto expansion to GradNorm solutions | GradNorm on MultiMNIST · MultiMNIST | hypervolume: 7.463 -> 7.628 | Helped |
| Apply Pareto expansion to Uncertainty solutions | Uncertainty on MultiMNIST · MultiMNIST | hypervolume: 7.615 -> 7.756 | Helped |
| Apply Pareto expansion to MGDA solutions | MGDA on MultiMNIST · MultiMNIST | hypervolume: 7.831 -> 7.896 | Helped |
| Apply Pareto expansion to WeightedSum solutions | WeightedSum on MultiMNIST · MultiMNIST | hypervolume: 8.019 -> 8.034 | Helped |
| Apply Pareto expansion to ParetoMTL solutions | ParetoMTL on MultiMNIST · MultiMNIST | hypervolume: 8.025 -> 8.046 | Helped |
| Apply Pareto expansion to WeightedSum solutions | WeightedSum on UCI Census-Income · UCI Census-Income | hypervolume: 5.685 -> 5.725 | Helped |
| Apply Pareto expansion to ParetoMTL solutions | ParetoMTL on UCI Census-Income · UCI Census-Income | hypervolume: 5.642 -> 5.675 | Helped |
| Apply Pareto expansion to ParetoMTL solutions | ParetoMTL on UTKFace · UTKFace | hypervolume: 2.257 -> 5.030 | Helped |
| Use tangent directions rather than WeightedSum expansion | Pareto expansion · MultiMNIST, UTKFace, and other tested datasets | Pareto-front dominance | Helped |
| Use tangent directions rather than WeightedSum expansion | Pareto expansion · UCI Census-Income | Pareto-front dominance | No clear effect |
| Set MINRES iteration limit to k=50 rather than k=20 (ablation) | Pareto expansion on MultiMNIST · MultiMNIST | hypervolume: 7.731 -> 7.734 | Helped |
| Increase MINRES iteration limit from k=50 to k=100 (ablation) | Pareto expansion on MultiMNIST · MultiMNIST | hypervolume: 7.734 -> 7.727 | Hurt |
Closest papers by meaning
- Efficient Pareto Manifold Learning with Low-Rank Structure (ICML 2024 · similarity 0.88)
- Improving Pareto Set Learning for Expensive Multi-objective Optimization via Stein Variational Hypernetworks (AAAI 2025 · similarity 0.86)
- Pareto Low-Rank Adapters: Efficient Multi-Task Learning with Preferences (ICLR 2025 · similarity 0.86)
- PSMGD: Periodic Stochastic Multi-Gradient Descent for Fast Multi-Objective Optimization (AAAI 2025 · similarity 0.86)
- Interactive Hyperparameter Optimization in Multi-Objective Problems via Preference Learning (AAAI 2024 · similarity 0.86)
- Quantifying Task Priority for Multi-Task Optimization (CVPR 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.