Transfer learning from language models to image caption generators: Better models may not transfer better.
Marc TantiAlbert GattKenneth P. CamilleriPublished in: CoRR (2019)
Keyphrases
- transfer learning
- language model
- probabilistic model
- knowledge transfer
- cross domain
- statistical language models
- learning tasks
- language modeling
- language modelling
- image retrieval
- reinforcement learning
- previously learned
- machine learning algorithms
- transfer knowledge
- image content
- n gram
- smoothing methods
- labeled data
- multi task
- text classification
- image segmentation
- cross lingual
- structure learning
- domain adaptation
- retrieval model
- relevance model
- semi supervised learning
- active learning
- similarity measure
- machine learning
- transferring knowledge
- text categorization
- query expansion
- image classification
- text mining
- collaborative filtering
- target domain
- visual features
- learning algorithm