Expected Likelihood: Not a Good Metric for Gaussian Embeddings.
Yiding TianLin QiuKewei TuYong YuPublished in: TAAI (2017)
Keyphrases
- maximum likelihood
- expected values
- distance measure
- metric space
- vector space
- euclidean space
- euclidean distance
- gaussian mixture model
- covariance matrices
- dimensionality reduction
- mixture of gaussians
- likelihood model
- neural network
- gaussian mixture
- posterior probability
- distance metric
- high dimensional data
- high dimensional