Multi-Stage Pre-training for Low-Resource Domain Adaptation.
Rong ZhangRevanth Gangi ReddyMd. Arafat SultanVittorio CastelliAnthony FerrittoRadu FlorianEfsun Sarioglu KayiSalim RoukosAvirup SilTodd WardPublished in: EMNLP (1) (2020)
Keyphrases
- multistage
- domain adaptation
- covariate shift
- single stage
- cross domain
- dynamic programming
- labeled data
- semi supervised
- multiple sources
- transfer learning
- lot sizing
- document classification
- test data
- manifold alignment
- test set
- sentiment classification
- semi supervised learning
- training set
- optimal policy
- training examples
- training process
- high dimensional
- data sets
- unlabeled data
- k nearest neighbor
- target domain
- model selection
- learning algorithm
- information retrieval