A multi-stage attention fusion and 1x1 convolution fusion model improves response generation for audio-visual scene-aware dialogue compared to the released baseline.
M.; Parikh, D.; and Batra, D
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Reactive Multi-Stage Feature Fusion for Multimodal Dialogue Modeling
A multi-stage attention fusion and 1x1 convolution fusion model improves response generation for audio-visual scene-aware dialogue compared to the released baseline.