WIMMC:加权的增量式极大边界准则算法

一些经典降维算法并不是最优的降维策略,它们不再适用于流形式且大尺度的Web文本数据,因此提出了一种加权的增量式有监督的降维算法,称为加权的增量式极大边界准则(Weighted Incremental Maximum Margin Criterion,WIMMC)。WIMMC通过加权得到比传统算法更好的结果,而且可以增量地有监督地处理大尺度的Web文本数据。给出了算法的收敛性证明和一些实验,并从实验结果可以看出,通过WIMMC降维之后的分类效果比其他降维算法更有效。     

塔里木盆地西南缘构造带的地球物理特征、构造及其演化

本文根据塔里木盆地西南缘构造带的重力异常，大地叫磁测深以及天然地震波速和航磁异常等地球物理资料，并结合该区的构造地质特征，将研究区划发为西昆仑叠瓦逆冲推覆带，推体前缘断褶带和外缘坳陷带。     

基于边界鉴别分析的递归维数约简算法

提出一种基于边界鉴别分析的递归维数约简算法.该算法把已求取边界鉴别向量正交于待求超平面法向量作为支持向量机(SVM)优化问题新的约束条件;然后对改进SVM进行递归求解,得到正交边界鉴别向量基;最后将数据样本在正交边界鉴别向量上投影实现维数约简.该算法不仅克服了现有维数约简算法难以支持小样本数据集、受数据样本分布影响等问题,而且抽取的特征向量具有更优的分类性能.仿真实验说明了算法的有效性.     

Effects of probe support on the stall characteristics of a low-speed axial compressor

In order to investigate the effects of probe support on the stall characteristics of micro compressors, an experiment was carried out on a large-scale low-speed research compressor according to the principle of geometric similarity. A cylindrical probe support intruding to 50% blade span was mounted at 50% chord upstream from the rotor blade leading edge. The static pressure rise characteristic of the compressor is measured, with and without the probe support respectively. The dynamic compressor behavior from pre-stall to full stall was also measured. The results indicate that the stability margin of the compressor is lowered after installing the probe support. The stall inception is aroused by modal wave disturbances. The disturbances developed into two stall cells smoothly before installing the probe support, while the disturbances first developed into a single stall cell then splitting into two stall cells after installing the probe support. The presence of probe support lowers the initial intensity of the rotating stall of the compressor, while it doesn't alter the intensity of the rotation stall after the compressor enters into full stall.     

Mean-reversion in income over feed cost margins: Evidence and implications for managing margin risk by US dairy producers

With the increased volatility of feed prices, dairy farm managers are no longer concerned with managing only milk price volatility, but are considering the adoption of risk management programs that address income over feed cost (IOFC) margin risk. Successful margin risk management should be founded on an understanding of the behavior of IOFC margins. To that end, we have constructed forward IOFC margins using Class III milk, corn, and soybean meal futures prices. We focus on the characteristics of the term structure of forward IOFC margins, that is, the sequence of forward margins for consecutive calendar months, all observed on the same trading day. What is apparent from the shapes of these term structures is that both in times when margins were exceptionally high and in times when they were disastrously low, market participants expected that a reversal back to average margin levels would not come quickly, but rather would take up to 9 mo. Slopes of the forward margin term structure before and after most of the major swings in IOFC indicate these shocks were mostly unanticipated, whereas the time needed for recovery to normal margin levels was successfully predicted. This suggests that IOFC margins may exhibit slow mean-reverting, rather than predictable cyclical behavior, as is often suggested in the popular press. This finding can be exploited to design a successful catastrophic risk management program by initiating protection at 9 to 12 mo before futures contract maturity. As a case study, we analyzed risk management strategies for managing IOFC margins that used Livestock Gross Margin for Dairy Cattle insurance contracts and created 2 farm profiles. The first one represents dairy farms that grow most of their feed, whereas the second profile is designed to capture the risk exposure of dairy farms that purchase all their dairy herd, dry cow, and heifer feed. Our case study of this program encompasses the 2009 period, which was characterized by exceptionally poor IOFC margin conditions. We analyzed the dynamics of realized IOFC margins in 2009 under 4 different risk management strategies and found that optimal strategies that were founded on the principles delineated above succeeded in reducing the decline in IOFC margins in 2009 by 93% for the Home-Feed profile and by 47% for the Market-Feed profile, and they performed substantially better than alternative strategies suggested by earlier literature.     

准噶尔盆地西北缘稠油油藏蒸汽吞吐后期提高开发效果的方法

准噶尔盆地西北缘浅层稠油油藏自1984年进行蒸汽吞吐试验以来,已陆续动用了六区、九区、红浅区等三个油田18个层块,截止1996年底共动用含油面积41.9Km2,地质储量9086×104t。通过近十年的开采,大部分区块已到了吞吐末期,吞吐阶段累积产油1308×104t,油汽比0.33,采出程度约20%左右,动用程度很低。本文就是利用油藏工程及数模方法,提出了在吞吐后期转换开采方式、加密调整、优化调整注采参数、改善采油工艺技术、层位上返、注添加剂等提高蒸汽吞吐后期开发效果的方法,通过这些方法,使稠油老区产量持续在稳定水平。     

基于间隔损失和L_1范数调节的特征选择方法研究

特征评价和选择是机器学习和模式识别的重要步骤。为了获得稀疏特征子集,结合间隔损失评估策略和L1范数调节技术来获得一种有效的特征选择方法（MLFWL-L1）,并将其应用到RBFSVM分类器。实验中,在UCI数据集上将提出的算法与Simba和ReliefF对比表明,验证所提出的算法是一种有效的特征选择方法。     

Kohonen SOM下的模糊神经电压稳定评估算法

为解决传统电压稳定评估计算方法存在的速度慢、精度低等问题,设计了一个多输入单输出的模糊神经网络。该网络通过Kohonen自组织映射聚类含SVC电力系统的有功和无功负荷来减少输入量个数,同时采用一个模糊输入的三层前馈神经网络进行训练,并评定电力系统的负荷能力裕度。经仿真验证,所设计的模糊神经网络计算精度高,有效改善了模糊神经网络的性能。     

Fuzzy Aggregation with Artificial Color filters

Artificial Color filters are designed to attenuate some pixels and pass others. The pass/attenuate decision is made on the basis of the learned association of spectral components with user-defined concepts. In earlier work, it has been shown that there are various ways to design Artificial Color filters using multiple user-designated classes and those filters are subjected to useful manipulations such as image processing and Boolean Aggregation.The Artificial Color filtering has always been binary. Therefore, the Boolean logic was the only choice for aggregating filters. This paper shows how to fuzzify Artificial Color filters. Fuzzy logic subsumes Boolean logic and can do so in many ways. Several different fuzzy T-norms are applied to Artificial Color filters to illustrate the richness in aggregation. Margin Setting, a supervised statistical pattern recognition method to train the filters, is very conservative in what is definitely assigned to a class (μ=1) while allowing a useful gradation of membership (μ?1) for other cases. A parametric exploration of these effects for an image is presented.