Research map / Learning theory and algorithms / Sample complexity
State-by-state Minimax Adaptive Estimation for Nonparametric Hidden {M}arkov Models
Luc Lehéricy
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 state-by-state selection instead of a single 10-fold cross-validated model | Spectral HMM estimator · Uniform emission density | convergence rate exponent: -0.434 ± 0.007 -> -0.500 ± 0.046 | Improved |
| use state-by-state selection instead of a single 10-fold cross-validated model | Spectral HMM estimator · Symmetrized Beta emission density | convergence rate exponent: -0.263 ± 0.011 -> -0.347 ± 0.007 | Improved |
| use state-by-state selection instead of a single 10-fold cross-validated model | Spectral HMM estimator · Beta emission density | convergence rate exponent: -0.377 ± 0.008 -> -0.470 ± 0.015 | Improved |
| use cross validation instead of state-by-state selection at small n | Spectral HMM estimator · Small sample size (n <= 350,000) | L2 error | Improved |
| use state-by-state selection instead of cross validation at large n | Spectral HMM estimator · Large sample size (n > 350,000) | L2 error: up to one order of magnitude | Improved |
| add POS variant of final estimator selection to Eachjump (ablation) | Eachjump calibration · Symmetrized Beta emission density | convergence rate exponent: -0.347 ± 0.007 -> -0.327 ± 0.008 | Hurt |
| add MAX variant to Jumpmax (ablation) | Jumpmax calibration · Symmetrized Beta emission density | convergence rate exponent: -0.349 ± 0.006 -> -0.404 ± 0.009 | Improved |
| replace Eachjump with Jumpmean calibration (ablation) | State-by-state selection · Uniform emission density | convergence rate exponent: -0.500 ± 0.046 -> -0.532 ± 0.048 | No clear effect |
| use MAX selection variant instead of standard selection (ablation) | State-by-state selection · Small number of observations | L2 error | Improved |
Closest papers by meaning
- Minimax Adaptive Estimation of Nonparametric Hidden Markov Models (JMLR 2016 · similarity 0.95)
- Learning HMMs with Nonparametric Emissions via Spectral Decompositions of Continuous Matrices (NeurIPS 2016 · similarity 0.89)
- Multiple Testing in Nonparametric Hidden Markov Models: An Empirical Bayes Approach (JMLR 2022 · similarity 0.87)
- Frontiers to the learning of nonparametric hidden Markov models (JMLR 2025 · similarity 0.87)
- Identifiability and Consistent Estimation of Nonparametric Translation Hidden Markov Models with General State Space (JMLR 2020 · similarity 0.86)
- Flat Minima in Linear Estimation and an Extended Gauss Markov Theorem (ICLR 2024 · 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.
