AI Paper Atlas

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

NeurIPS 2023 · Label-efficient and robust learning › Recognition and long-tail · approach: Contrastive representation theory · 2 citations (OpenAlex)

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 changedWhereMeasuredEffect
learn a Mahalanobis metric for the prediction loss instead of plain MSETaskMet · Cubic decision-focused problemdecision qualityImproved
TaskMet versus LODL (alpha=0)TaskMet · Cubic decision-focused problemdecision qualityNo clear effect
learned metric versus DFL with alpha=0TaskMet · Budget allocationdecision qualityHurt
learned metric versus DFL with alpha=0TaskMet · Budget allocationprediction error (MSE)Improved
task loss only (DFL alpha=0) instead of adding prediction loss (ablation)DFL · Portfolio optimizationdecision qualityHurt
MSE or DFL alpha=10 versus TaskMetTaskMet · Portfolio optimizationdecision qualityHurt
unconditional diagonal learned metric instead of MLETaskMet · Cartpole with distracting statesreturnImproved
MLE dynamics learning with distracting state dimensionsMLE · Cartpole with distracting statesreturnHurt
full conditional metric on under-parametrized model (3 hidden units) versus OMDTaskMet · Cartpole reduced capacityreturnImproved
TaskMet versus OMD on dynamics modelTaskMet · Cartpole reduced capacitymodel prediction MSEImproved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy