Research map / Probabilistic and scientific modeling / Variational and generative sampling
Geometric Dirichlet Means Algorithm for topic inference
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 tuned extension parameters (tGDM) instead of default extension | GDM · Simulation (varying N_m, M, eta) | minimum-matching Euclidean distance and perplexity | Helped |
| Replace RecoverKL with GDM as topic estimator | NIPS corpus · NIPS K=5 to 20 | perplexity: GDM 763-1269 vs RecoverKL 1378-1586 | Helped |
| Use weighted k-means with document lengths as weights (weighted GDM vs unweighted) (ablation) | GDM · Documents of varying size (50 to 1500) | perplexity and matching distance | Helped |
| Use tuned extension parameters with weighted GDM relative to default | GDM · Simulation with tuning of extension parameters for K=4 example | qualitative fit to Gibbs sampling | No clear effect |
| Use document-length-independent GDM in short documents (N_m=50) vs long | GDM · Simulation N_m=50, M up to 15000 | performance | No clear effect |
| Use anchor words inserted manually for RecoverKL | RecoverKL · Simulation with eta<=0.3 vs eta>=0.5, N_m=50 | topic estimation accuracy | Mixed |
| Use nGDM (nonparametric, penalized weighted k-means) vs KL DP-means | nGDM · Simulation varying alpha 0.01 to 2 | perplexity | Mixed |
Closest papers by meaning
- Scalable inference of topic evolution via models for latent geometric structures (NeurIPS 2019 · similarity 0.88)
- OTLDA: A Geometry-aware Optimal Transport Approach for Topic Modeling (NeurIPS 2020 · similarity 0.87)
- Conic Scan-and-Cover algorithms for nonparametric topic modeling (NeurIPS 2017 · similarity 0.87)
- A Reduction for Efficient LDA Topic Reconstruction (NeurIPS 2018 · similarity 0.87)
- Online Bayesian Moment Matching for Topic Modeling with Unknown Number of Topics (NeurIPS 2016 · similarity 0.86)
- Autoencoding Variational Inference For Topic Models (ICLR 2017 · similarity 0.86)
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.