ID 原文 译文
57888 采用向量化的克拉美罗下界作为布设优化策略，通过功效系数法优化布设策略，并采用天牛须搜索算法实现布设的快速、精确收敛. The algorithm adopts Cramér-Ｒao lower bound vector as the optimization strategy， and the efficiency coefficient method to optimize the layout strategy and then the algorithm uses beetle an- tennae search algorithm to achieve rapid deployment of anchor node.
57889 仿真结果表明: The simulations show that the pro- posed algorithm reduces the low bound of localization performance by 38.
57890 与传统区域定位误差均界评价指标相比，所提算法使得 99.74% 的区域内定位性能极限提升了约 38.79% ; 79% in 99. The simulations show that the pro- posed algorithm reduces the low bound of localization performance by 38. 79% in 99.74% of the regions，
57891 在 25 锚节点布设场景下，与遗传算法相比，布设后区域内定位性能极限大致相同，但搜索时间降低了约 64.2% . and in the 25-anchor node layout scenario，compared with the genetic algorithm，the low bound of locali- zation performance is almost same，but the search time is reduced by about 64. 2% .
57892 机器阅读理解中存在无法仅从给定文档中获取问题答案的特殊情况，为此，基于语义冲突检测的机器阅读理解网络( SCDNet) 提出应通过检测问题与文档内容之间的语义分歧来识别这种情况. Machine reading comprehension ( MＲC) with unanswerable questions is challenging to the field of natural language processing research. Unlike previous work which ignores the mechanism of an- swerable and unanswerable，the semantic conflicts detection-based MＲC network ( SCDNet) was pro- posed aiming at detections of no-answer ( NA) questions through semantic conflicts detection network.
57893 经分析发现，文档无法为问题提供答案的根本原因主要分为两类:一是文档中不包含问题所需的语义信息;二是二者包含的语义成分之间存在分歧. The basic idea is that if the given question is unanswerable，there exists semantic absence or conflicts be- tween the question and the reference passages.
57894 据此推断，可以通过检测文档语义信息是否全面涵盖问题所需的信息来识别问题是否可由文档信息给出回答. Therefore，SCDNet predicts the NA probability by chec- king whether the passage covers the integral semantics of the question.
57895 此外，通过在损失函数中加入答案文本长度惩罚项，网络优化目标函数更接近评测指标，系统性能得到提升. Besides，in order to extract the exact answer from the passage，SCDNet is applied an answer length penalty in the loss function，which helps the learning objective to be more consistent with the evaluation metrics.
57896 网络模型使用联合训练模型建模无答案的问题识别与答案抽取 2 个子任务，并使用端到端的方式训练. SCDNet packs the NA question predictor and the answer extractor in a joint model and is trained in an end-to-end manner.
57897 实验结果证明，其对无答案问题类别预测的正确率超过了性能先进的基线模型 SAN2.0，在 SQuAD2.0 数据集上取得了72.43 的 F1 值和 76.96 的无答案问题识别正确率. Ex- periments show that SCDNet performs better than some strong baseline models，and achieve an F1 score of 72. 43 and 76. 96 NA accuracy on SQuAD 2. 0 dataset.