基于深度学习的牙刷姿态识别方法

针对牙刷分拣中的定位问题,在确定牙刷位置的基础上采用深度学习实现牙刷姿态识别.对牙刷图像进行去噪增强,通过阈值分割提取感兴趣区域,计算图像的几何矩获得牙刷的方向角和外接矩形,以外接矩形的中心作为牙刷位置.用矩形框内的牙刷图像训练残差网络模型,当模型正确率达到要求时保存该模型,用于判断图像中牙刷的姿态.测试结果表明,该方法可以快速准确地实现牙刷的位置确定与姿态识别,为机器人分拣提供牙刷位姿信息.     

基于深度学习的中医古文献临床经验抽取

中医古文献蕴藏着丰富的临床经验，是古代中医在行医过程中对临床诊疗的经验性总结，体现了中医学形成和发展的理论框架和思想基础.然而这些宝贵的临床经验不仅量大，而且分散在不同的文献中，使得中医从业者手工很难快速全面地获取它们，文献检索工具也只能提供文档级别的信息筛选，无法为这种细粒度的信息获取提供支持.此外，古汉语相对于现代汉语的不同特点也限制了主流文本分析工具的使用效果.为此本文提出面向临床经验获取的中医古文献信息抽取任务，用于识别古文献中描述临床经验的文本片段，手工标注了样本数据用于这种抽取模型的训练和测试，并设计了基于深度学习的序列标注器用于完成该任务.考虑到标注数据量小可能带来的过度拟合问题，本文引入对抗训练和虚拟对抗训练来增强模型的泛化能力.一系列充分的实验验证了模型的有效性，表明利用信息抽取技术从古文献获取中医临床经验具有可行性，为这一新的信息抽取任务提供了有希望的研究基线和可复用的标注数据集.     

A new twist to the No Miracles Argument for the success of science

J. D. Trout has recently developed a new defense of scientific realism, a new version of the No Miracles Argument. I critically evaluate Trout's novel defense of realism. I argue that Trout's argument for scientific realism and the related explanation for the success of science are self-defeating. In the process of arguing against the traditional realist strategies for explaining the success of science, he inadvertently undermines his own argument.     

Little-known and phylogenetically obscure South African estuarine microgastropods (Mollusca: Truncatelloidea) as living animals

Areas of the Knysna estuarine bay in the Western Cape are dominated by three endemic South African truncatelloid microgastropods, temporarily known as ‘Hydrobia’ knysnaensis (Krauss), ‘Assiminea’ capensis (Sowerby) and ‘Assiminea’ globulus Connolly. Although first described 80–170 years ago and present in abundance (up to 100,000 m^?2), they remain surrounded by confusion and still await taxonomic assignment, largely because they appear most atypical members of their groups by virtue of anatomy and/or biogeography and/or habitat. This study contributes in-life perspectives to morphological and phylogenetic analyses known to be on-going. At Knysna, they are syntopic: at least two occurring in >85% and all three in >40% of individual 0.0026 m² samples from their region of dominance. Nevertheless, they tend to greater abundance in divergent microhabitats; ‘A.’ globulus dominating higher tidal levels, and ‘A.’ capensis and ‘Hydrobia’ lower ones; the former especially unvegetated sediment, the latter, if anything, seagrass. Interspecific feeding interactions appear unlikely to be responsible for these patterns, other evidence suggesting that all are maintained below carrying capacity. Field biology of ‘H.’ knysnaensis generally appears equivalent to that of northern-hemisphere intertidal hydrobiids and that of ‘A.’ globulus is typically assimineid, albeit at atypically low shore height. Unlike assimineids, however, ‘A.’ capensis is truly aquatic. The success of these truncatelloids in unusual circumstances may be consequent on the absence from South Africa of other microgastropod groups that fill their niches elsewhere in the southern hemisphere.     

Constructing and evaluating core inflation measures from component‐level inflation data

This paper undertakes a comprehensive examination of 10 measures of core inflation and evaluates which measure produces the best forecast of headline inflation out‐of‐sample. We use the Personal Consumption Expenditure Price Index as our measure of inflation. We use two sets of components (17 and 50) of the Personal Consumption Expenditure Price Index to construct these core inflation measures and evaluate these measures at the three time horizons (6, 12 and 24 months) most relevant for monetary policy decisions. The best measure of core inflation for both sets of components and over all time horizons uses weights based on the first principal component of the disaggregated (component‐level) prices. Interestingly, the results vary by the number of components used; when more components are used the weights based on the persistence of each component is statistically equivalent to the weights generated by the first principal component. However, those forecasts using the persistence of 50 components are statistically worse than those generated using the first principal component of 17 components. The statistical superiority of the principal component method is due to the fact that it extracts (in the first principal component) the common source of variation in the component level prices that accurately describes trend inflation over the next 6–24 months.     

抗坏血酸/谷氨酸Maillard反应体系中间产物和褐变程度的研究

用磷酸缓冲溶液配制一系列不同质量分数的抗坏血酸/谷氨酸混合溶液pH8.0并在130℃下加热发生Maillard反应,对反应后体系的中间产物(A294nm)、褐变程度(A420nm)以及反应过程中抗坏血酸和谷氨酸的含量进行检测分析.结果显示,抗坏血酸通过热解产物与氨基酸发生Maillard反应,抗坏血酸质量分数的增大能促进反应体系的褐变,谷氨酸质量分数对反应体系褐变程度的影响不大.任意增大一种反应物的质量分数均可对无色中间产物的生成起促进作用,且抗坏血酸对中间产物生成的影响大于谷氨酸.     

氧化还原介体催化强化污染物厌氧降解研究进展

 由于厌氧生物处理技术具有产生剩余污泥少、可回收能源等优点而被广泛用于处理各种有机污染物，尤其在有毒、有害、难降解污染物的去除方面取得了良好的效果。然而，厌氧生物法的处理速率通常比较低，而氧化还原介体可通过自身不断的氧化和还原来传递电子，提高电子在氧化还原反应过程中的传递速率，从而促进污染物高效厌氧降解。醌类物质和腐殖酸是应用较多的氧化还原介体，在催化难降解污染物降解方面取得了一定效果。讨论了氧化还原介体的特点、作用机制，并总结了其对偶氮染料厌氧脱色、反硝化和多氯联苯厌氧降解的强化作用，提出了氧化还原介体未来的研究方向。     

基于回波预处理和相参积累的C&I干扰抑制算法

C&I(chopping and interleaving)干扰是一种针对线性调频(linear frequency modulation, LFM)雷达的典型干扰样式, 干扰子信号调频斜率与雷达发射信号相同, 利用信号处理工具分离真实回波与干扰信号难度较大。针对该问题, 以LFM相参雷达抗自卫式C&I干扰为背景, 提出基于回波预处理和相参积累的干扰抑制算法。根据估计的回波时延, 设置距离窗截取受干扰回波段, 在此基础上, 通过对回波预处理, 改变不同重复周期内假目标的快时间位置分布, 通过相参积累实现干扰抑制。仿真试验表明, 所提算法能够有效抑制强干扰背景下的C&I干扰, 干扰抑制后真实目标检测概率大幅提高, 虚假目标数量明显减少。     

丙酸盐对厌氧氨氧化除氮性能及群落结构的影响

为了探究有机碳对厌氧氨氧化的长期影响, 向厌氧氨氧化膜生物反应器中添加不同浓度的丙酸钠, 研究对反应器的除氮效能以及微生物的种群结构和功能变化。结果表明: 反应器主要的脱氮过程由CandidatusBrocadia完成, 当丙酸钠浓度为100 mg/L时, 反应器中由于异养细菌的生长, 可实现碳氮的同步去除, 平均总氮去除率可达91.9%。当丙酸钠浓度为200 mg/L 时, 对Candidatus Brocadia的抑制作用导致反应器除氮性能下降, Candidatus Brocadia的丰度降至41.2%, 总氮去除率降至 78.8%。在有机碳抑制作用解除后, 反应器的除氮性能恢复为86.8%, Candidatus Brocadia丰度增加到54.0%, 但群落多样性下降, 属水平的微生物组成有较大的改变。