查询词典 root mean square error

In order to decrease the error rate of speech recognition, speaker adaptation techniques are often used to adjust speaker-dependent acoustic models. MLLR (Maximum Likelihood Linear Regression) and MAP are two of the most popular techniques in recent years. MLLR uses the technique of regression trees. It calculates the transform matrix for each leaf node of the tree. This makes it possible to use fewer sentences to decrease the error rate of HMM-based speech recognition. However, while we examined the recognition result, we found that although the overall error rate decreased, but the error rate of certain confusable phones was higher. In order to solve this problem, we propose the use MPE (Minimum Phone Error Discriminative Training) to solve this problem.

在语音辨识技术的实际应用上，语者调适技术常被用做调整语者无关之声学模型，使其对於特定语者之辨识率有所提升，常见的语者调适技术有最大化相似度线性回归（Maximum Likelihood Linear Regression），其精神在於透过对语音辨识模型相近的mixture做群聚，再对各个群聚做调整，以期能达到利用少量语料，提升辨识率的目的，但其缺点为，在发音上相近的模型，其mixture原本就极为近似，若再分类为同一群聚做调整，容易因为使用者的发音习惯，而使模型偏向ㄓ或ㄗ，因此虽然整体的辨识率获得提升，但却造成混淆音的错误率上升。

The conditions to measure accurately and the principles to select parameters of the multi— probes EST method are proposed, and the theoretical research, computer imitation and the experiment on a machine tool prove that these principles and suggestion are necessary and effective. The influence of the initial value errors of the probes, the installation error of the probes system, and the remaining error caused by the radical error motion are analyzed quantitively on the measurement accuracy of the multi—probes EST method. The influence of the error caused by the probe's axes which is not in a plane and the perpendicularity error of the plane which the probe's axes is in to the stational axis are discussed on the measurement results.

结合理论分析，对这些EST方法进行了计算机仿真分析和研究，重点研究了这些方法的应用条件和误差分离精度，所得出的结论是：三测头法EST基本上能实现各项误差的完全分离，适于在位测量使用；二测头法在恰当地选择测头安装角的条件下，可以得到较高测量精度，适于大型零件圆度、圆柱度、径向圆跳动和全跳动的在位测量；对置两测头时域法可以在一定程度上改善单测头法测量精度，但对随机径向误差无能为力；对置两测头频域法存在严重的谐波损失。

A commutation error is then considered in addition to the conventional residual error to generate the innovative residual error. Such an error is applied into cost functions to derive finite impulse response and infinite impulse response filter-based adaptive algorithms for ANC applications, referred to as FxRLS/CE and FuRLS/FRE+CE algorithms, respectively. Convergence analyses based on Lyapunov stability criteria for time-varying discrete-time systems can be carried out for the FxLMS/CE, FxRLS/CE and FuRLS/ FRE+CE algorithms to ensure stability. Computer simulations and experiments demonstrate that the innovative residual error-based adaptive algorithms can free the restriction of the slow-adaptation assumption in the conventional ANC approaches.

本文中也定义一交换误差并且加入到传统残留误差行形成新残留误差来改善适应性演算法的收敛速度，使用新残留误差推导出了有限脉冲响应滤波器与无限脉冲响应滤波器之适应演算法的新适应性演算法分别为： FxLMS/CE、 FxRLS/CE 和 FuRLS/FRE+CE 演算法，并且以离散时间的 Lyapunov 稳定性准则来证明演算法的收敛性，电脑模拟与实验也显示新演算法免除以往滤波器的参数必须缓慢更新的条件。

In order to evaluate aiming error of missile on the warship or on the submarine, including the error of warship position, the error of reference azimuth and the error of transfer measurement given by the navigation system, and try to divided them from the general error, it is highly important that we estimate and evaluate missile system of mobile bedding launching.

为了评估从舰船／潜艇上发射导弹的瞄准误差，包括舰船导航系统给定的船位误差、基准方位误差及其传递测量误差，并试图把它们从总的瞄准误差中分离出来，这对于动基座发射导弹武器的瞄准精度的评价极为重要。

As compared with the traditional irrigation-traditional fertilization, the following results of the improved traditional irrigation-optimized fertilization were obtained: reduced tiller number per unit area, lower percentage of the tiller ear number/the tiller number, the decreased ear number per unit area, the greatly reduced grain number per ear (1999-2000) and like weight of 1, 000 grains; almost the same plant height; shorter base internode length, a bit more weight per unit internode length, longer internode length under ear; lower LAI values in middle and later stages; coincident process of dry matter accumulation from straight growth onwards, a little more dry matter accumulation amount from earing to maturity; shorter and thicker flag leaf, 2〓 leaf, 3〓 leaf and 4〓 leaf from top, reduced total area of these leave; bigger mean canopy leaf tilt angle, smaller light extinction coefficient , higher ratio of light penetration in the upper, middle and lower parts of the wheat canopy, and tighter plant type; almost identical dynamic change of flag leaf chlorophyll content; and almost the same root amount in 0-120cm soil layer, smaller rate of progressive decrease of the root length density and root weight density with soil getting deep and relatively increased root amount in deeper soil layers.

在传统灌溉-优化施肥条件下，冬小麦单位面积总分蘖数比传统水肥减少，成穗率降低，单位面积穗数有所减少，穗粒数、千粒重无显著性差异；株高基本无变化；基部节间长度缩短，充实度略微增高，穗下节间增长；中后期叶面积指数略低于传统水肥；起身至成熟干物质积累量动态变化过程与传统水肥十分接近，但抽穗至成熟形成的干物质量增加；旗叶、倒2叶倒3叶、倒4叶叶长及总叶面积有所减少，叶片相对增厚；冠层平均叶倾角大，消光系数小，冠层各层的透光率较高，株型紧凑；旗叶叶绿素含量动态变化与传统水肥十分接近；0-120cm土层总根量与传统水肥也差异不大，但根长密度和根重密度随土层深度递减速度较慢，根系在下层土壤的分布相对增多。

The error correction method for microwave measuring line is analyzed, and a new method is proposed which can realize the measurement of error network parameters for microwave measuring line. Then the least square method explicit solutions of error network are given, and the CAT of error network parameters is realized.

分析了微波测量线误差校正的方法，提出了仅用小反射滑动负载即可确定测量线误差网络参数的新方法，给出了测量线误差网络参数的最小二乘法显式解，并实现了测量线误差网络参数的CAT 。

this article discusses the integral theorem of mean the promoted question, mainly has two aspects: On the one hand in analyzes in the teaching material under the first integral theorem of mean condition, had proven lies between the value spot to have to be possible to obtain in the open-interval, further discusses this knot promotes to the generalized Riemann integral, and further proved the conclusion also establishes to the promoted first integral theorem of mean; Promotes on the one hand in addition the integral theorem of mean to in the curve and the curved surface, and has proven the curvilinear integral theorem of mean and the surface integral theorem of mean.

本文讨论积分中值定理的推广问题，主要有二个方面：一方面在分析教材中第一积分中值定理的条件下，证明了介值点必可在开区间内取得，进一步将这个结论推广到广义Riemann积分，并进一步证明结论对推广的第一积分中值定理也成立；另一方面，将积分中值定理推广到曲线和曲面中，并证明了曲线积分中值定理和曲面积分中值定理。

In three separate studies, the investigators evaluated 105 adults with type 1 diabetes (mean age, 37 years; mean duration of diabetes, 20 years), 36 adults with type 2 diabetes (mean age, 50 years; mean duration of diabetes, 10 years), and 91 adults with type 1 diabetes (mean age, 39 years; mean duration of diabetes, 20 years). During a four-week period, participants used a hand-held computer to complete 70 trials of cognitive-motor tasks and to enter their current blood glucose reading after each test. Hyperglycemia was defined as blood glucose level higher than 15 mmol/L.

在3个不同的试验中，研究人员评估105位第1型糖尿病患者（平均年龄为37岁；患糖尿病平均时间为20年），36位第2型糖尿病患者（平均年龄为50岁；患糖尿病平均时间为10年），以及91位第1型糖尿病患者（平均年龄为39岁；患糖尿病平均时间为20岁）；在4周期间，受试者使用一种掌上型电脑来完成70种认知运动任务的试验，以及在每次测试后，输入他们目前的血糖读数，血糖高於15 mmolL被定义为高血糖。

Results the mean time of removed t-tube in patients with biliary leak were (28.7±7.3) days,the mean serum albumin level were (33.4±2.5)g/l,the mean hemoglobin level (116.5±17.3)g/l;while the mean time of removed t-tube in patients without biliary leak were (24.3±6.5)days,the mean serum albumin level were (37.9±1.7)g/l,the mean hemoglobin level (143.4±12.7)g/l.conclusion dystrophia is the reason of most biliary leak after removed t-tube.

结果 拔t管后胆漏病人的平均拔管时间为(28.7±7.3)天，血浆白蛋白(33.4±2.5)g/l，血红蛋白(116.5±17.3)g/l；无胆漏病人t管拔除的平均天数是(24.3±6.5)天，血浆白蛋白(37.9±1.7)g/l，血红蛋白(143.4±12.7)g/l。结论 t型引流管拔除后胆漏的主要原因是病人的营养状态差。

According to the energy conservation theory, BOM and CSIM4 were coupled. The BOM has no treatment on transmission solar radiation, which is of great importance when the model is adapted to Arctic Ocean. So the treatment was introduced to BOM. Through numerical test on different lead albedos, it was found that sea ice thickness is not so sensitive to lead albedo, which may be contribute to the lead occupies little ratio within multiyear sea ice pack. The reason of summer over-melt of arctic sea ice is the NCEP reanalysis downward solar radiation being larger than its reality. Then the arctic sea ice climate variability was simulated. Results showed that: simulated ice thickness change is in accord with the submarine investigated mean sea-ice draft changes. Simulated annually maximum ice thickness along the Eurasian continental oceans are closely related to the observed ones. The long-term mean simulated ice motion has the same features of the SSM/I derived ice motion. Sea ice extents in differential sub-regions have same trends comparing to the satellite passive-microwave data derived ones. Simulated ice concentration is closely related to the observed in the Arctic sub-regions. Sea ice flux through the Fram Strait involves ice concentration, motion and thickness. It is a composite criterion for sea ice model evaluation. The simulated ice area and volume export through the strait accord with the satellite derived or statistically reconstructed ones.(5) The simulated ice thickness climate variability and mean sea surface current of the coupled model were analyzed, results showed: the total ice volume in the Arctic Ocean has a significant decreasing trend. The volume variability is of a 10-year timescale oscillation, with two major periods of 12-13a and 18-20a. Mean ice thickness in the arctic sub-seas has different tendencies. It has an increasing trend in the Barents-Kara Sea and Baffin Bay-Labrador Sea, and decreasing in the others. The characteristic time scale of 7-10a wherein the river discharges leads the Fram Strait ice volume export is about the period that river water takes to be conveyed across the Arctic Ocean.(6) Using the simulated ice distribution in the Arctic Ocean and China precipitation, air temperature and SST in tropical key regions, the climate teleconnection were studied. Result showed: When the mean sea ice thickness is large in the central Arctic Ocean and Chukchi-Beaufort Sea , and small in the Barents-Kara Sea and Baffin Bay-Labrador Sea , the precipitation in South China, Tibetan Plateau, and the north part of Northeastern China are always smaller than normal, and v. v. When the mean ice thickness is small in CA, BC, East Siberian Sea and Greenland-Iceland-Norwegian Sea , and large in BL, The air temperature in north-eastern China, the southern of Tibetan Plateau, and Hainan Island, are always lower than normal, and v. v. In addition, when the sea ice is thick in BC and BL, the SST is larger in the middle and eastern Pacific Ocean, and is smaller in the tropical Southeastern Indian Ocean.

由于BOM没有考虑透射太阳辐射的物理过程，研究表明透射太阳辐射对北冰洋的能量收支起到重要作用，因此在BOM模式中引入了对透射太阳辐射的处理；通过对不同水道反照率的数值试验表明海冰厚度对水道反照率的敏感性不强，可能与海冰区水道面积占的比率很小有关；而模式模拟的北极海冰夏季&过度融化&主要源于NCEP再分析资料提供了偏大的太阳短波辐射；对北极海冰的气候变率进行了模拟研究，结果表明：模拟的海冰厚度变化与潜艇探测的海冰吃深度变化具有一致性；模拟和观测的亚欧大陆沿海的年内最大海冰厚度有很好的相关；模拟的海冰移速与长期平均的卫星反演的海冰移速具有相同的速度分布特征；模拟的各个海区海冰面积的变化趋势与卫星反演资料分析的结果基本一致；模拟与观测的主要海洋分区的海冰密集度具有很好的相关：弗瑞姆海峡的海冰体积和面积的输送涉及到海冰密集度、厚度和移动速度，是判断模式模拟能力的一个综合的指标，模式模拟的结果与卫星反演或重建的面积输送、体积输送具有很好的一致性；(5)分析了模拟的北极海冰厚度的气候变率及气候平均表层海流场，结果表明：北极海冰的总体积有显著减少的趋势，北极海冰总体积的变化具有10a际尺度振荡的特点，存在18-20a和12-13a两个主周期；北极海冰的平均厚度在各个海区的变化趋势不同，在巴伦支—喀拉海和巴芬湾—拉布拉多海地区海冰厚度有显著的增加趋势，而其它海区存在减少的趋势；通过对模拟的气候平均表层海流的分析表明，北极河流流量超前弗瑞姆海峡海冰流量7-10年的特征时间尺度与表层海流的气候分布存在着必然联系：(6)利用模拟结果以及中国降水、气温和热带关键区SST资料，讨论了北极各海区海冰平均厚度与中国降水、气温以及热带关键区SST的关系，结果表明：在北极中心海区和楚科奇—波弗特海海冰厚度偏大，在巴伦支—喀拉海以及巴芬湾—拉布拉多海海冰厚度偏小，则中国降水在华南地区、青藏高原和东北北部降水偏少，反之相反；在北极中心海区、东西伯利亚海、楚科奇—波弗特海以及格陵兰海海冰厚度偏小，在巴芬湾—拉布拉多海海冰厚度偏大，则在中国东北地区、高原南部地区和海南岛附近气温偏低，反之相反；另外，北极楚科奇—波弗特海和巴芬湾—拉布拉多海海冰厚度偏大时，在热带中东太平洋海温偏高，而在热带东南印度洋海温偏低。

Summary In copyright infringement cases,if the alleged infringer uses or even plagiarizes the copyrighted works or parts thereof,which causes no substantially adverse effect on the normal exploitation of the work,nor does it inflict any substantial damages to the right owner,the de minis doctrine should be applied to exempt the liabilities of the defendant and seek the balance of interests between the right owner and the general public embodied by the copyright law.

在侵犯著作权案件中，如果被控侵权行为人虽未经著作权人的同意，使用甚至抄袭了受著作权保护的作品或作品的片段，但若情节轻微，未对该作品的正常使用产生任何实质不利的影响，亦未对权利人的权利造成实质性的损害，则应适用&不计琐细原则&，免除行为人的责任，以寻求在权利人利益与著作权法所体现的社会公共利益之间的平衡。

This new product will become our best seller.

这个新产品会成为我们最畅销的产品。