Visually Grounded Word Embeddings and Richer Visual Features for Improving Multimodal Neural Machine Translation.
Jean-Benoit DelbrouckStéphane DupontOmar SeddatiPublished in: CoRR (2017)
Keyphrases
- visual features
- machine translation
- word sense disambiguation
- statistical machine translation
- word level
- word alignment
- target language
- machine translation system
- image classification
- visual information
- keywords
- image retrieval
- source language
- parallel corpus
- natural language processing
- cross lingual
- low level
- visual content
- image search
- image annotation
- cross language information retrieval
- information extraction
- low level features
- image collections
- key frames
- vector space
- semantic features
- natural language
- semantic concepts
- query translation
- translation model
- co occurrence
- document images
- multi modal
- artificial intelligence
- computer vision
- multimedia
- feature extraction
- sentence level
- machine learning