A Practical Method for Self-adapting Gaussian Expectation Maximization
Explore this paper's citation graph
Summary
This paper proposes a new EM algorithm that makes use of a on-line variable number of mixture Gaussians components and introduces a measure of the similarities to decide when to merge components.
- Type
- article
- Published
- 2010-01-01
- Cited by
- 3
- References
- 11
- Access
- Open access
- OpenAlex
- https://openalex.org/W4484436
- Semantic Scholar
- https://api.semanticscholar.org/CorpusID:6498622
Keywords
Computer science, Maximization, Gaussian, Gaussian process, Mathematical optimization
References
- Stochastic Complexity in Statistical Inquiry Theory
- Akaike information criterion statistics
- On the generalized distance in statistics
- Finite Mixture Models
- Unsupervised Learning of Finite Mixture Models
- Akaike Information Criterion Statistics
- A Measurement of Overlap Rate Between Gaussiancomponents
- SMEM Algorithm for Mixture Models
- EM algorithms for Gaussian mixtures with split-and-merge operation
- Genetic-based EM algorithm for learning Gaussian mixture models
- Maximum likelihood estimation from incomplete data via the EM algorithm
- Akaike Information Criterion Statistics.
Cited by
Related papers
- Constrained information maximization
- Analysis of influence maximization in large-scale social networks
- On supporting the maximization postulate
- A Decomposition of the Downlink Utility Maximization Problem
- Profit rate maximization in multipass turning with constraints : a geometric programming approach
- Estimating extremes of combined two Gaussian and non-Gaussian response processes
- Gaussian approximations to conditional distributions for multi-Gaussian processes