Diverse dialogue generation with context dependent dynamic loss function.
Ayaka UeyamaYoshinobu KanoPublished in: COLING (2020)
Keyphrases
- loss function
- context dependent
- support vector
- natural language
- pairwise
- semantic level
- logistic regression
- learning to rank
- reproducing kernel hilbert space
- risk minimization
- convex loss functions
- empirical risk
- context free
- regularization term
- data mining
- boosting algorithms
- model selection
- stochastic gradient descent
- learning algorithm
- hinge loss
- information retrieval