Experiments with Random Projection
Sanjoy Dasgupta · arXiv (Cornell University) · 2013
Recent theoretical work has identified random projection as a promising dimensionality reduction technique for learning mixtures of Gausians. Here we summarize these results and illustrate them by a wide variety of experiments on synthetic and real data.
Cited by 306 works. doi:10.48550/arxiv.1301.3849 Open-access PDF
Authors
References
- An introduction to probability theory and its applications
- Pattern classification and scene analysis
- Support-Vector Networks
- Neural networks for pattern recognition
- Mixture Densities, Maximum Likelihood and the EM Algorithm
- Learning mixtures of Gaussians
- Asymptotics of Graphical Projection Pursuit
- An algorithmic theory of learning: Robust concepts and random projection
- Estimating Dependency Structure as a Hidden Variable