Research map / Learning theory and algorithms / Combinatorial and search optimization
Synthesizing Action Sequences for Modifying Model Decisions
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 |
|---|---|---|---|
| Combine search over action sequences with optimization of action parameters | German model · German Credit Data test set | instances solved | Helped |
| Combine search over action sequences with optimization of action parameters | Adult model · Adult Dataset test set | instances solved | Helped |
| Combine search over action sequences with optimization of action parameters | Fannie Mae model · Fannie Mae Loan Performance test set | instances solved | Helped |
| Use gradient-guided sequence scoring instead of breadth-first scoring (ablation) | Action-sequence search · German Credit Data, Adult Dataset, and Fannie Mae | optimization problems solved before finding a solution | Mixed |
| Use objective-value sequence scoring instead of breadth-first scoring (ablation) | Action-sequence search · German Credit Data, Adult Dataset, and Fannie Mae | optimization problems solved before finding a solution | Mixed |
| Use gradient-guided scoring rather than objective-value scoring (ablation) | Action-sequence search · Fannie Mae | optimization problems solved before finding a solution | Helped |
| Use objective-value scoring rather than gradient-guided scoring (ablation) | Action-sequence search · Adult Dataset | optimization problems solved before finding a solution | Helped |
| Add random perturbations to synthesized action parameters (ablation) | Synthesized sequences for the Adult model · Adult Dataset | success under parameter noise | Hurt |
| Use action sequences with up to 6 added strokes | Drawing Recognition model · Quick, Draw! cat sketches | instances solved | Helped |
Closest papers by meaning
- Verification of the Implicit World Model in a Generative Model via Adversarial Sequences (ICLR 2026 · similarity 0.85)
- AdvJudge-Zero: Binary Decision Flips in LLM-as-a-Judge via Adversarial Control Tokens (NeurIPS 2026 · similarity 0.83)
- AdvPrompter: Fast Adaptive Adversarial Prompting for LLMs (ICML 2025 · similarity 0.83)
- Enabling VLA Action Self-Verification via VLM Token Probability Bucketing (NeurIPS 2026 · similarity 0.83)
- Reverse Engineering Human Preferences with Reinforcement Learning (NeurIPS 2025 · similarity 0.83)
- VLA-ATTC: Adaptive Test-Time Compute for VLA Models with Relative Action Critic Model (ICML 2026 · 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.