Xiang Cheng, Yulan Hu, Xiangwen Zhang, Lu Xu, Lide Tan, Zheng Pan, Xin Li, Yong Liu. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Zhicong Li, Lingjie Jiang, Yulan Hu, Xingchen Zeng, Yixia Li, Xiangwen Zhang, Guanhua Chen, Zheng Pan, Xin Li, Yong Liu. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Prompting has recently been shown as a promising approach for applying pre-trained language models to perform downstream tasks.We present Multi-Stage Prompting, a simple and automatic approach for leveraging pre-trained language models to translation tasks.To better mitigate the discrepancy between pre-training and translation, MSP divides the translation process via pre-trained language models into multiple separate stages: the encoding stage, the re-encoding stage, and the decoding stage.During each stage, we independently apply different continuous prompts for allowing pretrained language models better shift to translation tasks.We conduct extensive experiments on three translation tasks.Experiments show that our method can significantly improve the translation performance of pre-trained language models.

Urban forests can maintain urban ecological balance and improve environmental quality, but it is difficult to identify such forests accurately due to its complex and fragmented features. This study aims to develop a deep-learning network to extract the urban forest spatial distribution from high-spatial resolution image, like from Chinese Gaofen-2 (GF-2) image. Based on the GF-2 surface reflectance image and urban forest samples known in prior, this study firstly create a U-Net network to train and generate a predictive model to identify the urban forest, and then used the trained model to get the spatial distribution of urban forest in the Beibei district. Results showed that the U-Net Network can predict urban forests distribution accurately and rapidly.

Taking the Dexing Copper Mine in Jiangxi Province, China as an study area, we used the long-term sequence summer Landsat images in 2002-2019 to investigate the variation of vegetation growth status and their ecological restoration effects. According to the specific situation of the study area, the Green-Red Normalized Difference Vegetation Index (GRNDVI) calculated from remote sensing data was used to analyse the growth of mine vegetation and the dynamic change in the whole mining area. Moreover, the annual growth changes were compared with normal vegetation growth. The CV method, Hurst method, and Sen+Mann-Kendall method were combined used to evaluate the intensity of vegetation growth, change patterns and change sustainability analysis to obtain the overall growth and change of vegetation and then predict the vegetation growth trend in the study area. The results show that this method can assess accurately the vegetation growth trend in the study area.

The dominant neural machine translation (NMT) models apply unified attentional encoder-decoder neural networks for translation. Traditionally, the NMT decoders adopt recurrent neural networks (RNNs) to perform translation in a left-to-right manner, leaving the target-side contexts generated from right to left unexploited during translation. In this paper, we equip the conventional attentional encoder-decoder NMT framework with a backward decoder, in order to explore bidirectional decoding for NMT. Attending to the hidden state sequence produced by the encoder, our backward decoder first learns to generate the target-side hidden state sequence from right to left. Then, the forward decoder performs translation in the forward direction, while in each translation prediction timestep, it simultaneously applies two attention models to consider the source-side and reverse target-side hidden states, respectively. With this new architecture, our model is able to fully exploit source- and target-side contexts to improve translation quality altogether. Experimental results on NIST Chinese-English and WMT English-German translation tasks demonstrate that our model achieves substantial improvements over the conventional NMT by 3.14 and 1.38 BLEU points, respectively. The source code of this work can be obtained from https://github.com/DeepLearnXMU/ABDNMT.