AI Paper Atlas

Research map / Large language models / Agents, code and math

BayesAgent: Bayesian Agentic Reasoning Under Uncertainty via Verbalized Probabilistic Graphical Modeling

Hengguan Huang, Xing Shen, Guang-Yuan Hao, Songtao Wang, Lingfa Meng, Dianbo Liu, David Alejandro Duchene, Hao Wang, Samir Bhatt

AAAI 2026 · Large language models › Agents, code and math · approach: Inductive and abstract reasoning evaluation · 5 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
add verbalized PGM latent-variable reasoning (N=3) instead of ChameleonChameleon agent (Llama3-8B-Instruct) · ScienceQAECE (x10^2): 9.62 -> 1.67Improved
add verbalized PGM latent-variable reasoning (N=3) instead of ChameleonChameleon agent · ScienceQAaccuracy: 85.29 -> 86.38Improved
add numerical Bayesian inference (Dirichlet posterior, learned lambda) to vPGMvPGM N=3 · ScienceQAECE (x10^2): 1.67 -> 1.05Improved
add numerical Bayesian inference to vPGMvPGM N=3 · ScienceQAaccuracy: 86.38 -> 86.38No clear effect
increase latent variables from 3 to 4 (ablation)vPGM · ScienceQAECE (x10^2): 1.67 -> 2.15Hurt
increase latent variables from 2 to 3 (ablation)vPGM · ScienceQAaccuracy: 85.49 -> 86.38Improved
use vPGM instead of GCoTGCoT · ChatCoach detectionBERTScore: 72.4 -> 76.3Improved
use vPGM instead of GCoTGCoT · ChatCoach correctionBLEU-2 / Rouge-L: BLEU-2 1.6 -> 1.7; BERTScore 65.4 -> 68.3; ROUGE-L 2.0 -> 2.0Mixed
use vPGM instead of Chameleon+ on misinformation inputsChameleon+ · A-OKVQA-noisyaccuracy: 59.04 -> 61.03Improved
add BayesVPGM inference on top of vPGMvPGM · A-OKVQA-noisyECE (x10^2): 10.54 -> 9.85Improved
latent variable Z2 flags rationale mismatchBayesVPGM · A-OKVQA-clean vs noisyP(Z2): 0.86 clean vs 0.42 noisyImproved
use latent variables when no misinformation is presentBayesVPGM · A-OKVQA-cleancalibrationHurt

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