Hao Wang, Linlong Xu, Heng Liu, Yangyang Liu, Xiaohu Zhao, Bo Zeng, Liangying Shao, Yichen Dong, Xinwei Wu, Jiang Zhou, Tianyu Dong, Xiangxiang Zeng, Longyue Wang, Weihua Luo. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Lei Yang, Leiyu Pan, Bojian Xiong, Renren Jin, Shaowei Zhang, Yue Chen, Ling Shi, Jiang Zhou, Junru Wu, Zhen Wang, Jianxiang Peng, Juesi Xiao, Tianyu Dong, Zhuowen Han, Zhuo Chen, Yuqi Ren, Deyi Xiong. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Jiang Zhou, Xiaohu Zhao, Xinwei Wu, Tianyu Dong, Hao Wang, Yangyang Liu, Heng Liu, Linlong Xu, Longyue Wang, Weihua Luo, Deyi Xiong. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

High-speed permanent magnet (HSPM) machines are pivotal in aerospace, industrial automation, and electric vehicles owing to their exceptional power density and compact structure. However, their design optimization faces significant challenges arising from the inherent complexity of multiphysics interactions, the computational cost of high-fidelity simulations, and the limitations of conventional optimization methods in handling conflicting objectives. To overcome these issues, this article proposes a novel collaborative framework that synergizes finite element analysis (FEA), interpretable machine learning Light Gradient Boosting Machine (LightGBM), and deep reinforcement learning (DRL) for efficient electromagnetic-thermal co-optimization of HSPM machines. First, construct a FEA dataset covering key structural variables and multiphysics field responses. A high-fidelity surrogate model (SM) based on the LightGBM was developed. The SHapley Additive exPlanations (SHAP)-partial dependence plots (PDP) interpretable analysis of this model revealed critical nonlinear mechanisms. This surrogate model is subsequently embedded within the Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm, forming a data-driven DRL optimization model that dynamically balances electromagnetic performance and thermal constraints. The proposed framework, termed MPOSM-TD3, is experimentally validated to simultaneously reduce the maximum winding temperature by 15.3$^\circ \text{C}$and improve operational efficiency by 0.94%.

Large language models (LLMs) have achieved remarkable progress in multilingual machine translation (MT), demonstrating strong performance even with limited parallel data.However, effectively fine-tuning LLMs for MT is challenging due to parameter interference, which arises from the conflicting demands of different language pairs and the risk of overwriting pre-trained knowledge.To address this issue, we propose MLAS-LoRA, a novel multiple language-aware LoRA knowledge transfer framework.MLAS-LoRA efficiently adapts LLMs to MT by selectively transferring knowledge from a large teacher to a small student model.Our approach first evaluates the awareness of neurons and extracts linguistic knowledge in the teacher model to both the general MT task and specific language pairs.We then propose a multiple language-specific LoRA architecture to inject the extracted knowledge into the student model.During fine-tuning, only the parameters of the relevant languagegeneral and language-specific LoRA modules are updated.Experimental results on diverse multilingual language pairs demonstrate that MLAS-LoRA significantly outperforms strong baselines by +1.7 BLEU on average, including standard fine-tuning and other parameterefficient methods.