Transformers can optimally learn regression mixture models.
Reese PathakRajat SenWeihao KongAbhimanyu DasPublished in: CoRR (2023)
Keyphrases
- mixture model
- model selection
- gaussian mixture model
- em algorithm
- density estimation
- probabilistic model
- generative model
- regression model
- mixture modeling
- unsupervised learning
- expectation maximization
- language model
- maximum likelihood
- automatic model selection
- probability density function
- cross validation
- mixture components
- pattern recognition
- image processing
- log linear models
- subspace analysis
- computer vision
- finite mixtures
- learning algorithm