Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Stopping Criterion Design for Recursive Bayesian Classification: Analysis and Decision Geometry
Aziz Koçanaoğulları, Murat Akçakaya, Deniz Erdogmus B
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 MP instead of Shannon entropy criterion (M3) with a disfavored prior | recursive Bayesian classification · 3-class toy, edge prior | accuracy among stops at sequence 2: 0.20 -> 0.67 | Improved |
| use MP instead of Renyi entropy alpha=0.2 (M4) | recursive Bayesian classification · 3-class toy, edge prior | p_stop: 1.00 -> 0.59 at sequence 5 | Improved |
| use MP instead of confidence threshold M1 at tau=0.8 | recursive Bayesian classification · 3-class toy, edge prior | p_stop and accuracy: 0.59/0.97 -> 0.59/0.97 | No clear effect |
| use MP instead of lowered confidence threshold M1-bar | recursive Bayesian classification · 10-class toy, edge prior | accuracy among stops at sequence 5: 0.87 -> 0.94 | Improved |
| use MP instead of M1 with a central prior | recursive Bayesian classification · 10-class toy, central prior | sequence reaching p_stop>=0.5: 15th -> 14th sequence | Improved |
| increase class count from 3 to 10 | entropy criterion M3 · toy examples | p_stop: 0.80 -> 1.00 at sequence 5 | Hurt |
| use MP instead of M1 in BCI typing with uniform prior | BCI typing · RSVP keyboard 100 letters | sequences needed: 1945 -> 1735 | Improved |
| use MP instead of M1 in BCI typing with a language model prior | BCI typing · RSVP keyboard 100 letters | sequences needed: 1728 -> 1580 | Improved |
| use MP instead of M1 in BCI typing | BCI typing · RSVP keyboard, uniform prior | typing accuracy (%): 94% -> 90% | Hurt |
| use KL divergence between posteriors (M5) instead of MP | BCI typing · RSVP keyboard, uniform prior | average extra sequences: +3482 sequences | Hurt |
Closest papers by meaning
- Recursive PAC-Bayes: A Frequentist Approach to Sequential Prior Updates with No Information Loss (NeurIPS 2024 · similarity 0.84)
- Optimal Strategies for Reject Option Classifiers (JMLR 2023 · similarity 0.83)
- On discriminative learning of prediction uncertainty (ICML 2019 · similarity 0.83)
- To Each Metric Its Decoding: Post-Hoc Optimal Decision Rules of Probabilistic Hierarchical Classifiers (ICML 2025 · similarity 0.82)
- Learning the Optimal Stopping for Early Classification within Finite Horizons via Sequential Probability Ratio Test (ICLR 2025 · similarity 0.82)
- Stop the Sampler! Classifier-Based Adaptive Stopping for Sampling Kernels (NeurIPS 2026 · similarity 0.82)
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.
