融合约束学习的图像字幕生成方法

目的 图像字幕生成是一个涉及计算机视觉和自然语言处理的热门研究领域，其目的是生成可以准确表达图片内容的句子。在已经提出的方法中，生成的句子存在描述不准确、缺乏连贯性的问题。为此，提出一种基于编码器-解码器框架和生成式对抗网络的融合训练新方法。通过对生成字幕整体和局部分别进行优化，提高生成句子的准确性和连贯性。方法 使用卷积神经网络作为编码器提取图像特征，并将得到的特征和图像对应的真实描述共同作为解码器的输入。使用长短时记忆网络作为解码器进行图像字幕生成。在字幕生成的每个时刻，分别使用真实描述和前一时刻生成的字幕作为下一时刻的输入，同时生成两组字幕。计算使用真实描述生成的字幕和真实描述本身之间的相似性，以及使用前一时刻的输出生成的字幕通过判别器得到的分数。将二者组合成一个新的融合优化函数指导生成器的训练。结果 在CUB-200数据集上，与未使用约束器的方法相比，本文方法在BLEU-4、BLEU-3、BLEI-2、BLEU-1、ROUGE-L和METEOR等6个评价指标上的得分分别提升了0.8%、1.2%、1.6%、0.9%、1.8%和1.0%。在Oxford-102数据集上，与未使用约束器的方法相比，本文方法在CIDEr、BLEU-4、BLEU-3、BLEU-2、BLEU-1、ROUGE-L和METEOR等7个评价指标上的得分分别提升了3.8%、1.5%、1.7%、1.4%、1.5%、0.5%和0.1%。在MSCOCO数据集上，本文方法在BLEU-2和BLEU-3两项评价指标上取得了最优值，分别为50.4%和36.8%。结论 本文方法将图像字幕中单词前后的使用关系纳入考虑范围，并使用约束器对字幕局部信息进行优化，有效解决了之前方法生成的字幕准确度和连贯度不高的问题，可以很好地用于图像理解和图像字幕生成。     

公益型矿业科研院所绩效考核机制探索

绩效考核一直是公益型科研院所管理领域的重点和难点。本文以公益型矿业科研院所为研究对象，系统地梳理绩效考核政策的历史沿革，分析国有科研院所“课题组承包制”绩效评价方法，总结“课题组承包制”绩效考核体系的优缺点；探索“以项目为中心”绩效考核体系，从考核基本原则、指标设计、评定方法与标准以及考核组织与实施方法等四个维度建立公益型矿业科研院所绩效考核评价方法；最后，提出了未来企业绩效考核机制的改进建议。为类似的矿业领域科研院所的人力资源管理模式提供了有益参考和借鉴。     

学生参与电气人才创新能力培养的探索

基于“以学生为本”的OBE教育理念，本文对电气人才创新能力培养进行了探索。整个过程将专业课程体系契合学业规划以明确学生学习目标；学生参与理论、实践教学和平台搭建，在夯实其理论知识的同时丰富教学资源；在创新考核中采用“螺旋递进式”的师生互动模式，全程培养学生自主的创新能力。多年实践探索表明学生在课程实践、学科竞赛、论文发表和申报专利等方面的创新能力得到显著提升。     

基于动态行为学习的空中目标识别方法

针对空中移动目标识别问题,提出了一种基于动态行为学习的识别方法。首先,从数据源中提取目标要素,经过标注形成目标数据特征集;其次,基于应用场景进行特征组合,形成特征模型要素;然后,基于深度学习算法进行模型训练,得到评估最优的模型参数;最后,利用训练后的模型进行目标识别处理。通过数值仿真验证,该方法能够有效地识别无业务规则的目标身份。     

Convolutional and generative adversarial neural networks in manufacturing

Manufacturing is undergoing transformation driven by the developments in process technology, information technology, and data science. A future manufacturing enterprise will be highly digital. This will create opportunities for machine learning algorithms to generate predictive models across the enterprise in the spirit of the digital twin concept. Convolutional and generative adversarial neural networks have received some attention of the manufacturing research community. Representative research and applications of the two machine learning concepts in manufacturing are presented. Advantages and limitations of each neural network are discussed. The paper might be helpful in identifying research gaps, inspire machine learning research in new manufacturing domains, contribute to the development of successful neural network architectures, and getting deeper insights into the manufacturing data.     

以用户为中心的可见光通信协作传输中训练资源分配算法

以用户为中心的可见光通信协作传输是近年来出现的新架构,这导致虚拟小区之间出现重叠。为避免导频污染问题,每个虚拟小区中的光接入点(AP)或者虚拟小区中选择相同AP的用户发送的训练序列应该是正交的。针对可见光通信中以用户为中心的协作网络,研究训练资源的正交分配问题,提出了一种新的导频分配算法,联合导频分配和用户选择问题,以期最大限度地增加虚拟小区内可被接入的用户数。分析和仿真结果表明,该导频分配方案可以有效改善导频污染问题,提高训练资源利用率,并且相比已有的导频分配方案,性能有所改进。     

Image Super-Resolution Based on Generative Adversarial Networks: A Brief Review

Single image super resolution (SISR) is an important research content in the field of computer vision and image processing. With the rapid development of deep neural networks, different image super-resolution models have emerged. Compared to some traditional SISR methods, deep learning-based methods can complete the superresolution tasks through a single image. In addition, compared with the SISR methods using traditional convolutional neural networks, SISR based on generative adversarial networks (GAN) has achieved the most advanced visual performance. In this review, we first explore the challenges faced by SISR and introduce some common datasets and evaluation metrics. Then, we review the improved network structures and loss functions of GAN-based perceptual SISR. Subsequently, the advantages and disadvantages of different networks are analyzed by multiple comparative experiments. Finally, we summarize the paper and look forward to the future development trends of GAN-based perceptual SISR.     

Adaptive retrieval practice with multiple-choice questions in the university classroom

Retrieval practice promotes retention more than restudying (i.e., the testing effect) and is applied to many educational settings. However, little research has investigated means to enhance this effect in educational settings. Theoretical accounts assume retrieval practice to be the most effective whenever retrieval is difficult but successful. Therefore, we developed a novel retrieval practice procedure, which adapts to learners' abilities and can be applied irrespective of learning content. This adaptive procedure aims to make retrieval gradually easier whenever students provide an incorrect answer. In a field experiment, students read book chapters as part of a weekly university course. In three consecutive weeks, they then practiced reading assignments by (a) adaptive testing, (b) non-adaptive testing and (c) restudy. In Week 4, a surprise criterial test took place. Restudy outperformed both testing conditions, whereas adaptive testing performed equally well as non-adaptive testing. However, exploratory analyses revealed that with increasing retention intervals, the superiority of restudy disappeared. Furthermore, whenever participants fully read the assignments and retention intervals increased, adaptive testing outperformed non-adaptive testing. In sum, adaptive retrieval practice did not prove to be generally superior, but retention interval and students' preparation for class might be conditions rendering adaptive retrieval useful in educational settings.     

装配式装修人才培养模式的研究与实践

装配式装修作为一个全新的课题被提出,其课题价值受到愈来愈多人的重视。其所体现的特征如标准化设计、工厂化生产、装配(式)化施工、一体化装修、(智能)信息化管理等概念,也被认为符合整个建筑业装配化发展的大趋势。本文探讨了校企合作在推动产业与学科建设、培养适应产业发展需求人才等方面的重要作用,为装配式装修人才培养模式提供参考。     

一种半监督对抗鲁棒模型无关元学习方法

元学习期望训练所得的元模型在学习到的“元知识”基础上利用来自新任务的少量标注样本，仅通过较少的梯度下降步骤微调模型就能够快速适应该任务。但是，由于缺乏训练样本，元学习算法在元训练期间对现有任务过度训练时所得的分类器决策边界不够准确，不合理的决策边界使得元模型更容易受到微小对抗扰动的影响，导致元模型在新任务上的鲁棒性能降低。提出一种半监督对抗鲁棒模型无关元学习（semi-ARMAML）方法，在目标函数中分别引入半监督的对抗鲁棒正则项和基于信息熵的任务无偏正则项，以此优化决策边界，其中对抗鲁棒正则项的计算允许未标注样本包含未见过类样本，从而使得元模型能更好地适应真实应用场景，降低对输入扰动的敏感性，提高对抗鲁棒性。实验结果表明，相比ADML、R-MAML-TRADES等当下主流的对抗元学习方法，semi-ARMAML方法在干净样本上准确率较高，在MiniImageNet数据集的5-way 1-shot与5-way 5-shot任务上对抗鲁棒性能分别约提升1.8%和2.7%，在CIFAR-FS数据集上分别约提升5.2%和8.1%。