Research map / Learning theory and algorithms / Prediction and decision losses
Be Inaccurate but Don’t Be Indecisive: How Error Distribution Can Affect User Experience
Rafael R. Padovani, Lucas N. Ferreira, Levi H. S. Lelis
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 |
|---|---|---|---|
| use unanimity-vote ensemble EC(4) instead of single NS classifier | Naive Bayes music selector · CotW 9 episodes | average SM: 40 -> 5 | Improved |
| use ensemble EC(4) instead of NS | Naive Bayes music selector · CotW 9 episodes | average accuracy (%): 64 -> 65 | No clear effect |
| enlarge ensemble from 2 to 4 classifiers (add NM and NHM) (ablation) | EC · CotW 9 episodes | average SM: 9 -> 5 | Improved |
| enlarge ensemble from 2 to 4 classifiers (ablation) | EC · CotW 9 episodes | average accuracy (%): 65 -> 65 | No clear effect |
| add MI-based feature selection (NS to NHS) (ablation) | Naive Bayes · CotW 9 episodes | average SM: 40 -> 36 | No clear effect |
| select window size by lowest SM instead of highest accuracy (NS to NM) (ablation) | Naive Bayes · CotW 9 episodes | average SM: 40 -> 36 | No clear effect |
| feature selection plus SM-based window selection (NHM vs NS) (ablation) | Naive Bayes · CotW 9 episodes | average SM: 40 -> 35 | No clear effect |
| use EC(4) instead of always-Suspenseful baseline | EC(4) · CotW 9 episodes | average accuracy (%): 42 -> 65 | Improved |
| EC(4) instead of NS when NS misclassifications are sparse (V1, V5) | EC · user study V1 and V5 | % preference: 57.5 vs 10.0; 47.5 vs 22.5 | Improved |
| EC picks decisively wrong Agitated song for Calm scene | EC · user study V2 | % preference: 57.5 vs 12.5 | Improved |
| EC picks Agitated song for sneaking Suspenseful scene | EC · user study V3 | % preference: 35.0 vs 37.5 | Hurt |
| EC with lower accuracy but lower SM than NS | EC · user study V4 | % preference: 35.0 vs 37.5 | Hurt |
Closest papers by meaning
- Do LLMs estimate uncertainty well in instruction-following? (ICLR 2025 · similarity 0.81)
- On Online Experimentation without Device Identifiers (ICML 2024 · similarity 0.81)
- The Burden of Interactive Alignment with Inconsistent Preferences (NeurIPS 2025 · similarity 0.81)
- Quantifying Uncertainty in Error Consistency: Towards Reliable Behavioral Comparison of Classifiers (NeurIPS 2025 · similarity 0.80)
- Expectation Alignment: Handling Reward Misspecification in the Presence of Expectation Mismatch (NeurIPS 2024 · similarity 0.80)
- On Targeted Manipulation and Deception when Optimizing LLMs for User Feedback (ICLR 2025 · similarity 0.80)
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.
