Congratulations! NLP's one paper is accepted by AAAI 2023! The full name of AAAI 2023 is the Thirty-Seventh AAAI Conference on Artificial Intelligence, which is one of the top conferences in artificial intelligence. It is supported annually by the AAAI, the Association for the Advancement of Artificial Intelligence. AAAI 2023 will be held in Washington, D.C., USA from February 7 to February 14, 2023.
The accepted paper is summarized as follows:
- Rephrasing the Reference for Non-Autoregressive Machine Translation (Chenze Shao, Jinchao Zhang, Jie Zhou, Yang Feng)
- AAAI Main Conference, long paper
Abstract: Non-autoregressive neural machine translation (NAT) significantly improves the decoding speed by generating all tokens in parallel. The major challenge that limits the NAT performance is the multi-modality problem that there may exist multiple possible translations of a source sentence. Therefore, the reference sentence may be inappropriate for the training when the NAT output is closer to other translations. In response to this problem, we introduce a rephraser to provide a better training target for NAT by rephrasing the reference sentence according to the NAT output. As we train NAT based on the rephraser output rather than the reference sentence, the rephraser output should fit well with the NAT output and not deviate too far from the reference, which can be quantified as reward functions and optimized by reinforcement learning. Experiments on major WMT benchmarks and NAT baselines show that our approach consistently improves the translation quality of NAT. Specifically, our best variant achieves comparable performance to the autoregressive Transformer, while being 14.7 times more efficient in inference.