Research map / Large language models / Agents, code and math
STAND: Self-Aware Precondition Induction for Interactive Task Learning
Daniel Weitekamp, Glen R Smith, Ken Koedinger, Christopher J. MacLellan
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 |
|---|---|---|---|
| add hierarchical shrinkage to STAND (ablation) | STAND · Synthetic precondition data | final holdout accuracy %: 90.0 -> 92.8 | Improved |
| add hierarchical shrinkage to decision tree (ablation) | Decision Tree · Synthetic precondition data | final holdout accuracy %: 88.6 -> 87.81 | No clear effect |
| add hierarchical shrinkage to STAND (ablation) | STAND · Synthetic precondition data | false positive reoccurrence rate %: 2.6 -> 2.2 | Improved |
| add hierarchical shrinkage to STAND (ablation) | STAND · Synthetic precondition data | Brier Score: 0.0046 -> 0.0029 | Improved |
| add hierarchical shrinkage to STAND (ablation) | STAND · Synthetic precondition data | Expected Calibration Error: 0.147 -> 0.123 | Improved |
| add hierarchical shrinkage to decision tree (ablation) | Decision Tree · Synthetic precondition data | false positive reoccurrence rate %: 7.3 -> 8.9 | Hurt |
| use STAND space-of-classifiers instead of decision tree | decision tree learner · Synthetic precondition data | false positive reoccurrence rate %: 7.3 -> 2.6 | Improved |
| use STAND instead of ensembles for certainty scores | STAND · Synthetic precondition data | productive monotonicity: 60-70% vs not much higher than 50% | Improved |
| use STAND with hierarchical shrinkage for mislabel detection with few examples | STAND (h.s.) · Mislabeled sample detection, 20 examples | top-1 detection rate %: 18.0 -> 37.8 (vs XGBoost) | Improved |
| use STAND on large noisy UCI data | STAND · 6 UCI datasets (mean) | test accuracy %: 89.18 vs 89.12 (tree), 89.67 (RF) | No clear effect |
| use STAND with hierarchical shrinkage on real ITL tasks | STAND (h.s.) · Multi-column Addition (AI2T) | final holdout accuracy %: 95.4 -> 99.43 | Improved |
Later papers that compared with this method
1 later accepted paper reports STAND as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Efficient Training-Free Multi-Token Prediction via Embedding-Space Probing (ICML 2026 · better on 8 of 8 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Generalizing Verifiable Instruction Following (NeurIPS 2025 · similarity 0.85)
- RuleReasoner: Reinforced Rule-based Reasoning via Domain-aware Dynamic Sampling (ICLR 2026 · similarity 0.84)
- SwS: Self-aware Weakness-driven Problem Synthesis in Reinforcement Learning for LLM Reasoning (NeurIPS 2025 · similarity 0.84)
- J1: Incentivizing Thinking in LLM-as-a-Judge via Reinforcement Learning (ICLR 2026 · similarity 0.84)
- SRPO: Self-Reflective Policy Optimization for Long-Horizon Reasoning (ICML 2026 · similarity 0.84)
- Skill-SD: Skill-Conditioned Self-Distillation for Multi-turn LLM Agents (NeurIPS 2026 · similarity 0.84)
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.
