Metric Learning Loss Functions to Reduce Domain Mismatch in the x-Vector Space for Language Recognition.
Raphaël DuroselleDenis JouvetIrina IllinaPublished in: INTERSPEECH (2020)
Keyphrases
- vector space
- loss function
- metric learning
- pairwise
- distance metric
- distance metric learning
- pairwise constraints
- distance measure
- reproducing kernel hilbert space
- similarity search
- distance function
- semi supervised
- dimensionality reduction
- low dimensional
- learning tasks
- euclidean space
- feature vectors
- multi task
- mahalanobis distance
- vector space model
- feature extraction
- pattern recognition
- support vector
- semi supervised learning
- kernel matrix
- riemannian manifolds
- computer vision
- feature space
- domain adaptation
- machine learning
- learning problems
- markov random field
- neural network