Research map / Learning theory and algorithms / Combinatorial and search optimization
Teaching the Old Dog New Tricks: Supervised Learning with Constraints
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 |
|---|---|---|---|
| Use Moving Targets with a neural network under balance constraints | Neural Network · Redwine | constraint violation: $1^{+} \pm .5$ -> $.0 \pm .05$ | Helped |
| Use Moving Targets with a neural network under balance constraints | Neural Network · Whitewine | constraint violation: $1^{+} \pm 1$ -> $.0 \pm .2$ | Helped |
| Use Moving Targets with a neural network under balance constraints | Neural Network · Shuttle | constraint violation: $1^{+} \pm 0$ -> $.6 \pm .3$ | Helped |
| Use β=.05 instead of β=.1 (ablation) | Moving Targets (NN) · Iris | accuracy: $\mathbf{.997} \pm .004$ -> $\mathbf{.997} \pm .004$ | No clear effect |
| Use β=.05 instead of β=.1 (ablation) | Moving Targets (NN) · Redwine | accuracy: $.506 \pm .006$ -> $\mathbf{.511} \pm .009$ | No clear effect |
| Use β=.01 instead of β=.1 (ablation) | Moving Targets (NN) · Whitewine | accuracy: $.439 \pm .009$ -> $\mathbf{.446} \pm .006$ | Mixed |
| Use β=.01 instead of β=.1 (ablation) | Moving Targets (NN) · Shuttle | accuracy: $.375 \pm .007$ -> $.39 \pm .04$ | No clear effect |
| Set α=.1 and β=.01 instead of α=1 and β=.1 (ablation) | Moving Targets (NN) · Dota2 | accuracy: $.66 \pm .01$ -> $\mathbf{.672} \pm .004$ | Mixed |
| Use β=.05 instead of β=.1 (ablation) | Moving Targets (NN) · Adult | accuracy: $.841 \pm .006$ -> $\mathbf{.86} \pm .02$ | No clear effect |
| Set α=.1 and β=.01 instead of α=1 and β=.1 (ablation) | Moving Targets (NN) · Crime | R2: $.48 \pm .03$ -> $.45 \pm .05$ | No clear effect |
| Use direct label preprocessing rather than iterative Moving Targets | NN_pp · Redwine | constraint violation: $.0 \pm .05$ -> $1^{+} \pm .3$ | Hurt |
| Use Moving Targets rather than regularization-based methods | Constraint-aware learning · Whitewine | accuracy: $.15 \pm .03$ -> $.439 \pm .009$ | Helped |
Closest papers by meaning
- Self-Supervised Transformers as Iterative Solution Improvers for Constraint Satisfaction (ICML 2025 · similarity 0.86)
- DeepSaDe: Learning Neural Networks That Guarantee Domain Constraint Satisfaction (AAAI 2024 · similarity 0.86)
- Near-Optimal Solutions of Constrained Learning Problems (ICLR 2024 · similarity 0.85)
- Learning to Learn in Interactive Constraint Acquisition (AAAI 2024 · similarity 0.84)
- Online Learning in CMDPs: Handling Stochastic and Adversarial Constraints (ICML 2024 · similarity 0.83)
- A Unifying Post-Processing Framework for Multi-Objective Learn-to-Defer Problems (NeurIPS 2024 · 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.