Research map / Label-efficient and robust learning / Recognition and long-tail
TaskMet: Task-driven Metric Learning for Model Learning
Dishank Bansal, Ricky T. Q. Chen, Mustafa Mukadam, Brandon Amos
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| learn a Mahalanobis metric for the prediction loss instead of plain MSE | TaskMet · Cubic decision-focused problem | decision quality | Improved |
| TaskMet versus LODL (alpha=0) | TaskMet · Cubic decision-focused problem | decision quality | No clear effect |
| learned metric versus DFL with alpha=0 | TaskMet · Budget allocation | decision quality | Hurt |
| learned metric versus DFL with alpha=0 | TaskMet · Budget allocation | prediction error (MSE) | Improved |
| task loss only (DFL alpha=0) instead of adding prediction loss (ablation) | DFL · Portfolio optimization | decision quality | Hurt |
| MSE or DFL alpha=10 versus TaskMet | TaskMet · Portfolio optimization | decision quality | Hurt |
| unconditional diagonal learned metric instead of MLE | TaskMet · Cartpole with distracting states | return | Improved |
| MLE dynamics learning with distracting state dimensions | MLE · Cartpole with distracting states | return | Hurt |
| full conditional metric on under-parametrized model (3 hidden units) versus OMD | TaskMet · Cartpole reduced capacity | return | Improved |
| TaskMet versus OMD on dynamics model | TaskMet · Cartpole reduced capacity | model prediction MSE | Improved |
Closest papers by meaning
- Leaving the Nest: Going beyond Local Loss Functions for Predict-Then-Optimize (AAAI 2024 · similarity 0.83)
- Bridging Risk Approximation Gaps in Model Predictive Task Sampling via In-Context Modeling (NeurIPS 2026 · similarity 0.83)
- Task-Aware Virtual Training: Enhancing Generalization in Meta-Reinforcement Learning for Out-of-Distribution Tasks (ICML 2025 · similarity 0.83)
- HarmonyDream: Task Harmonization Inside World Models (ICML 2024 · similarity 0.83)
- Representation Learning Enables Scalable Multitask Deep Reinforcement Learning (NeurIPS 2026 · similarity 0.83)
- Why Has Predicting Downstream Capabilities of Frontier AI Models with Scale Remained Elusive? (ICML 2025 · similarity 0.83)
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.
