Research map / Probabilistic and scientific modeling / Variational and generative sampling
Automated Extraction of Mutual Independence Patterns Using Bayesian Comparison of Partition Models
Guillaume Marrelec, Alain Giron
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 |
|---|---|---|---|
| add 2-way stochastic hierarchical clustering steps to Gibbs sampling (ablation) | Gibbs sampler · fMRI (82 regions) | L1 distance between runs: 1.996 -> 1.323 | Improved |
| add parallel tempering (L=7) to Gibbs sampling (ablation) | Gibbs sampler · fMRI (82 regions) | L1 distance between runs: 1.996 -> 0.275 | Improved |
| add parallel tempering to Gibbs+2wSHC (ablation) | Gibbs+2wSHC sampler · fMRI (82 regions) | L1 distance between runs: 1.323 -> 0.335 | Improved |
| add Gibbs steps to 2wSHC (ablation) | 2wSHC sampler · fMRI (82 regions) | L1 distance between runs: 1.982 -> 1.323 | Improved |
| add 2wSHC steps to Gibbs+PT (ablation) | Gibbs+PT sampler · fMRI (82 regions) | L1 distance between runs and failures: 0.275 -> 0.335 | Hurt |
| use Gibbs alone versus 2wSHC alone (ablation) | Gibbs vs 2wSHC · fMRI (82 regions) | between-chain heterogeneity: 3/2 | No clear effect |
| reduce chain length of Gibbs+2wSHC+PT from 10^5 to 10^4 samples (ablation) | Gibbs+2wSHC+PT · fMRI (82 regions) | failed runs out of 10: 10 -> 1 | Improved |
| use full Bayesian posterior (BayesOptim) versus BIC approximation (ablation) | mutual independence extraction · HIV data (D=6) | posterior probability of top model 12356|4: 0.852 vs 0.912 | No clear effect |
| use BayesCorr hyperparameters instead of BayesOptim (ablation) | mutual independence extraction · HIV data (D=6) | posterior probability of top model 12356|4: 0.852 -> 0.648 | Mixed |
| increase number of blocks K in the simulated partition | BayesOptim · simulation D=6, Gaussian data | posterior probability of true model | Hurt |
| increase sample size | BayesOptim · simulation D=6, Gaussian data | posterior probability of true model | Improved |
| replace Gaussian data with Student-t data of low degrees of freedom | BayesOptim · simulation D=6 | variability of posterior quantities | Hurt |
Closest papers by meaning
- Measuring Dependence Powerfully and Equitably (JMLR 2016 · similarity 0.86)
- A Bayesian Contiguous Partitioning Method for Learning Clustered Latent Variables (JMLR 2021 · similarity 0.84)
- Measuring Conditional Independence by Independent Residuals: Theoretical Results and Application in Causal Discovery (AAAI 2018 · similarity 0.84)
- Bivariate Causal Discovery using Bayesian Model Selection (ICML 2024 · similarity 0.83)
- Discovering Mixtures of Structural Causal Models from Time Series Data (ICML 2024 · similarity 0.83)
- Global-Local Dirichlet Processes for Clustering Grouped Data in the Presence of Group-Specific Idiosyncratic Variables (ICML 2025 · 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.
