AI Paper Atlas

Research map / Probabilistic and scientific modeling / Variational and generative sampling

Joint Bayesian Inference of Graphical Structure and Parameters with a Single Generative Flow Network

Tristan Deleu, Mizu Nishikawa-Toomey, Jithendaraa Subramanian, Nikolay Malkin, Laurent Charlin, Yoshua Bengio

NeurIPS 2023 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Flow matching and flow sampling · 4 citations (OpenAlex)

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 changedWhereMeasuredEffect
use JSP-GFN (diag) instead of best baseline BCD Netsjoint posterior approximation · d=5 linear Gaussianedge features RMSE: 0.215 ± 0.055 -> 0.018 ± 0.005Improved
use JSP-GFN (diag) instead of VBGjoint posterior approximation · d=5 linear Gaussianedge features Pearson's r: 0.816 ± 0.064 -> 0.998 ± 0.001Improved
use JSP-GFN (diag) instead of DiBSparameter posterior · d=5 linear Gaussiancross-entropy: 9.49 ± 7.34 × 103 -> −4.91 ± 0.51 × 100Improved
full-covariance Gaussian instead of diagonal in JSP-GFN (ablation)JSP-GFN · d=5 linear Gaussianedge features RMSE: 0.018 ± 0.005 -> 0.019 ± 0.007No clear effect
use JSP-GFN instead of MCMC methods on flow cytometryJSP-GFN · Flow cytometryheld-out NLL: 2.578 × 104 vs 2.679 × 104Mixed
use JSP-GFN instead of MH-MC3 on gene expressionJSP-GFN · Gene expression d=61held-out NLL: 1.073 × 106 -> 2.651 × 105Improved
use JSP-GFN instead of Gibbs-MC3 on flow cytometryJSP-GFN · Flow cytometryheld-out NLL: 3.490 × 105 -> 2.679 × 104Improved
compare JSP-GFN against baselines on linear Gaussian networksJSP-GFN · Linear Gaussian d=20held-out NLLImproved
compare JSP-GFN against baselines on non-linear Gaussian networksJSP-GFN · Non-linear Gaussian d=20held-out NLLNo clear effect
use whole dataset (N=4200) instead of subsample (N=100) for reward (ablation)JSP-GFN · Flow cytometryslope of log-prob vs log-reward: 1.0272 -> 0.7817Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy