Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Sparse Markov Models for High-dimensional Inference
Guilherme Ost, Daniel Y. Takahashi
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 FSC(5) lag selection instead of no selection (Naive) at d=5 | transition probability estimator · MTD simulation Exp2 (lags 1,5; d=5) | empirical std of p(0|0^d) estimate: 0.0340 -> 0.0222 (n=4096), 0.0252 -> 0.0129 (n=8192) | Improved |
| use FSC(5) instead of PCP for lag selection at d=5 | transition probability estimator · MTD simulation Exp2 (lags 1,5; d=5) | empirical std of p(0|0^d) estimate: 0.0242 -> 0.0222 (n=4096), 0.0131 -> 0.0129 (n=8192) | No clear effect |
| use FSC(3) instead of best subset selection BSS(2) for lag recovery | MTD lag selection · MTD simulation Exp1 (lags 1,8; d=8) | probability of correct lag selection: 0.41 -> 0.53 (n=5.10^3), 0.87 -> 1 (n=5.10^5) | Improved |
| use CTF(3) instead of FSC(3) for lag selection at moderate sample size | MTD lag selection · MTD simulation Exp1 (lags 1,8; d=8) | probability of correct lag selection: 0.81 -> 0.99 (n=10^4), 1 -> 1 (n=5.10^5) | No clear effect |
| use FS(2) with known number of lags instead of FSC(5) | MTD lag selection · MTD simulation Exp1 (lags 1,15; d=15) | probability of correct lag selection: 0.51 -> 0.66 (n=10^3), 0.82 -> 0.92 (n=5.10^3) | Improved |
| increase order to d=n/4 with relevant lag n/5, FSC(5) versus FS(2) | FSC(5) · MTD simulation Exp1 (lags 1,n/5; d=n/4) | probability of correct lag selection: 0.46 -> 1 (n=10^4 to 5.10^4), FS(2) 0.96 -> 1 | Mixed |
| apply FSC, CTF and VLMC to Canberra rain data (n=1000) | lag detection · Canberra daily rain | detected relevant lags: lags {1, 62, 330} versus order 1 with VLMC | Improved |
Closest papers by meaning
- Markov Latent Feature Models (ICML 2016 · similarity 0.87)
- Exploiting locality in high-dimensional Factorial hidden Markov models (ICML 2023 · similarity 0.86)
- Scalable Inference of Sparsely-changing Gaussian Markov Random Fields (NeurIPS 2021 · similarity 0.86)
- Variational Inference in high-dimensional linear regression (JMLR 2022 · similarity 0.85)
- Beyond normality: Learning sparse probabilistic graphical models in the non-Gaussian setting (NeurIPS 2017 · similarity 0.85)
- Scaling Factorial Hidden Markov Models: Stochastic Variational Inference without Messages (NeurIPS 2016 · 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.
