Lattice-Based Risk Minimization Training for Unsupervised Language Model Adaptation.
Akio KobayashiTakahiro OkuShinichi HommaToru ImaiSeiichi NakagawaPublished in: INTERSPEECH (2011)
Keyphrases
- training data
- risk minimization
- supervised learning
- generalization error
- training process
- training set
- training samples
- empirical risk
- unsupervised learning
- loss function
- data sets
- supervised training
- programming language
- learning algorithm
- natural language
- uniform convergence
- line search
- hinge loss
- deep architectures
- training algorithm
- machine learning