AI Paper Atlas

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

Principled Selection of Hyperparameters in the Latent Dirichlet Allocation Model

Clint P. George, Hani Doss

JMLR 2018 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Sequential Monte Carlo for inference · 59 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 ST estimator I-hat_st (with extra stored averages) instead of I-tilde_st for posterior probability estimation (ablation)Serial tempering on Wikipedia Cats corpus · Wikipedia Cats corpus, nu(||theta7-theta8||<=.07)variability across independent runs: two independent I-hat estimates close; two I-tilde estimates notImproved
Use serial tempering M-hat instead of M-tilde for marginal likelihood estimation (ablation)Serial tempering · Synthetic and Wikipedia corporaestimator varianceImproved
Use Serial tempering instead of VI-EM to estimate hyperparameterLDA hyperparameter estimation · Synthetic corpora K=4,8; D=20,40,100accuracy of h estimate vs true argmax: VI-EM estimates far from true value and depend on starting valuesImproved
Use Serial tempering instead of Gibbs-EMLDA hyperparameter estimation · Synthetic corpora K=4,8; D=20,40,100accuracy of h estimate: Gibbs-EM points far from inside 95% confidence ellipseImproved
Use Gibbs-EM instead of VI-EMEM-based hyperparameter estimation · Synthetic corporaaccuracy of h estimateImproved
Double Gibbs chain length each EM iteration (m_n = 2^(6+n)) (ablation)Gibbs-EM · Synthetic corporaaccuracy of h estimateNo clear effect
Increase number of documents D from 20 to 100VI-EM and Gibbs-EM · Synthetic corporaaccuracy of h estimateImproved
Use Gibbs-EM instead of VI-EM for speedhyperparameter estimation · Synthetic corporacomputation time: time for Gibbs-EM about 55 times the time for VI-EMHurt
Use Serial tempering instead of Gibbs-EM for speedhyperparameter estimation · Synthetic corporacomputation time: time for ST about seven times the time for Gibbs-EMHurt
Increase number of documents in synthetic corpusSerial tempering argmax · Synthetic corporacloseness of h-hat to h_trueImproved
Use Augmented Collapsed Gibbs Sampler vs Grouped Gibbs Sampler (ablation)Serial tempering chain · Wikipedia Cats corpusestimatesNo clear effect
Use collapsed Gibbs sampler instead of grouped Gibbs sampler (ablation)LDA posterior sampling · LDA mixing ratemixing rateImproved

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