DOLDA: a regularized supervised topic model for high-dimensional multi-class regression

DOI10.1007/S00180-019-00891-1MaRDI QIDQ2184404zbMATH OpenOpenAlexFDO

Authors Leif Jonsson, Mattias Villani, Måns Magnusson

Publication date 28 May 2020

Published in Computational Statistics (Search for Journal in Brave)

Copyright license Creative Commons Attribution 4.0 International

Full work available at URL https://arxiv.org/abs/1602.00260

zbMATH Keywords

horseshoe priorinterpretable modelstext classificationlatent Dirichlet allocationdiagonal orthant probit model

Mathematics Subject Classification ID

Computational methods for problems pertaining to statistics (62-08) Bayesian inference (62F15) Classification and discrimination; cluster analysis (statistical aspects) (62H30) Learning and adaptive systems in artificial intelligence (68T05)

Abstract: Generating user interpretable multi-class predictions in data rich environments with many classes and explanatory covariates is a daunting task. We introduce Diagonal Orthant Latent Dirichlet Allocation (DOLDA), a supervised topic model for multi-class classification that can handle both many classes as well as many covariates. To handle many classes we use the recently proposed Diagonal Orthant (DO) probit model (Johndrow et al., 2013) together with an efficient Horseshoe prior for variable selection/shrinkage (Carvalho et al., 2010). We propose a computationally efficient parallel Gibbs sampler for the new model. An important advantage of DOLDA is that learned topics are directly connected to individual classes without the need for a reference class. We evaluate the model's predictive accuracy on two datasets and demonstrate DOLDA's advantage in interpreting the generated predictions.

Recommendations

• MedLDA: maximum margin supervised topic models
• 10.1162/jmlr.2003.3.4-5.993
• Gibbs max-margin topic models with data augmentation
• A correlated topic model of science
• Principled selection of hyperparameters in the latent Dirichlet allocation model

Cites work

• 10.1162/jmlr.2003.3.4-5.993
• A Bayesian analysis of the multinomial probit model using marginal data augmentation
• Bayesian Analysis of Binary and Polychotomous Response Data
• Bayesian Inference for Logistic Models Using Pólya–Gamma Latent Variables
• Bayesian linear regression with sparse priors
• Distributed algorithms for topic models
• Gibbs Sampling for Bayesian Non-Conjugate and Hierarchical Models by Using Auxiliary Variables
• MedLDA: maximum margin supervised topic models
• Sparse Partially Collapsed MCMC for Parallel Inference in Topic Models
• Statistical topic models for multi-label document classification
• The horseshoe estimator for sparse signals
• Tree ensembles with rule structured horseshoe regularization

Cited in

(3)

• MedLDA: maximum margin supervised topic models
• Lasso meets horseshoe: a survey
• Robust supervised topic models under label noise

Describes a project that uses

Uses Software

• Medlda
• tokenizers
• horserule

This page was built for publication: DOLDA: a regularized supervised topic model for high-dimensional multi-class regression

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2184404)