E-resources
-
Chu, Jiqun; Lin, Zuoquan
Proceedings of the 2022 5th International Conference on Algorithms, Computing and Artificial Intelligence, 12/2022Conference Proceeding
Encoder-decoder model encodes input sentences to hidden representations and decodes the representations to the output in unidirectional way. We introduce a bidirectional encoder decoder model that adds a reverse decoder-encoder for the feedback from the output to the input. We implement the bidirectional encoder decoder model based on the transformer. The decoder-encoder attention mechanism enhances the representations via the interaction between encoder and decoder. Experiment results demonstrate the benefit of bidirectional model. And the visualization results of bidirectional representations reveal that our model learn an entangled representation of both encoder and decoder.
