Early Gains Matter: A Case for Preferring Generative over Discriminative Crowdsourcing Models.
Paul FeltKevin BlackEric K. RinggerKevin D. SeppiRobbie HaertelPublished in: HLT-NAACL (2015)
Keyphrases
- model selection
- unsupervised learning
- discriminatively trained
- face recognition
- discriminative methods
- data sets
- classification models
- statistical models
- machine learning algorithms
- discriminative models
- bayesian network classifiers
- computational models
- experimental data
- generative model
- semi supervised
- high dimensional
- information retrieval