Additive Gaussian processes

We introduce a Gaussian process model of functions which are additive. An additive function is one which decomposes into a sum of low-dimensional functions, each depending on only a subset of the input variables. Additive GPs generalize both Generalized Additive Models, and the standard GP models which use squared-exponential kernels. Hyperparameter learning in this model can be seen as Bayesian Hierarchical Kernel Learning (HKL). We introduce an expressive but tractable parameterization of the kernel function, which allows efficient evaluation of all input interaction terms, whose number is exponential in the input dimension. The additional structure discoverable by this model results in increased interpretability, as well as state-of-the-art predictive power in regression tasks.

Parametric guidance for turbulent noise reduction from poroelastic trailing edges and owls

The interaction of a turbulent eddy with a semi-infinite, poroelastic edge is examined with respect to the effects of both elasticity and porosity on the efficiency of aerodynamic noise generation. The edge is modelled as a thin plate poroelastic plate, which is known to admit fifth-, sixth-, and seventh-power noise dependences on a characteristic velocity U of the turbulent eddy. The associated acoustic scattering problem is solved using the Wiener-Hopf technique for the case of constant plate properties. For the special cases of porous-rigid and impermeable-elastic plate conditions, asymptotic analysis of the Wiener- Hopf kernel function furnishes the parameter groups and their ranges where U5, U6, and U7 behaviours are expected to occur. Results from this analysis attempt to help guide the search for passive edge treatments to reduce trailing-edge noise that are inspired by the wing features of silently flying owls. Furthermore, the appropriateness of the present model to the owl noise problem is discussed with respect to the acoustic frequencies of interest, wing chord-lengths, and foraging behaviour across a representative set of owl species.

Kernelized log linear models for continuous speech recognition

Large margin criteria and discriminative models are two effective improvements for HMM-based speech recognition. This paper proposed a large margin trained log linear model with kernels for CSR. To avoid explicitly computing in the high dimensional feature space and to achieve the nonlinear decision boundaries, a kernel based training and decoding framework is proposed in this work. To make the system robust to noise a kernel adaptation scheme is also presented. Previous work in this area is extended in two directions. First, most kernels for CSR focus on measuring the similarity between two observation sequences. The proposed joint kernels defined a similarity between two observation-label sequence pairs on the sentence level. Second, this paper addresses how to efficiently employ kernels in large margin training and decoding with lattices. To the best of our knowledge, this is the first attempt at using large margin kernel-based log linear models for CSR. The model is evaluated on a noise corrupted continuous digit task: AURORA 2.0. © 2013 IEEE.

Student-t processes as alternatives to Gaussian processes

We investigate the Student-t process as an alternative to the Gaussian process as a non-parametric prior over functions. We derive closed form expressions for the marginal likelihood and predictive distribution of a Student-t process, by integrating away an inverse Wishart process prior over the co-variance kernel of a Gaussian process model. We show surprising equivalences between different hierarchical Gaussian process models leading to Student-t processes, and derive a new sampling scheme for the inverse Wishart process, which helps elucidate these equivalences. Overall, we show that a Student-t process can retain the attractive properties of a Gaussian process - a nonparamet-ric representation, analytic marginal and predictive distributions, and easy model selection through covariance kernels - but has enhanced flexibility, and predictive covariances that, unlike a Gaussian process, explicitly depend on the values of training observations. We verify empirically that a Student-t process is especially useful in situations where there are changes in covariance structure, or in applications such as Bayesian optimization, where accurate predictive covariances are critical for good performance. These advantages come at no additional computational cost over Gaussian processes.

Geometrical learning, descriptive geometry, and biomimetic pattern recognition

Studies on learning problems from geometry perspective have attracted an ever increasing attention in machine learning, leaded by achievements on information geometry. This paper proposes a different geometrical learning from the perspective of high-dimensional descriptive geometry. Geometrical properties of high-dimensional structures underlying a set of samples are learned via successive projections from the higher dimension to the lower dimension until two-dimensional Euclidean plane, under guidance of the established properties and theorems in high-dimensional descriptive geometry. Specifically, we introduce a hyper sausage like geometry shape for learning samples and provides a geometrical learning algorithm for specifying the hyper sausage shapes, which is then applied to biomimetic pattern recognition. Experimental results are presented to show that the proposed approach outperforms three types of support vector machines with either a three degree polynomial kernel or a radial basis function kernel, especially in the cases of high-dimensional samples of a finite size. (c) 2005 Elsevier B.V. All rights reserved.

Object-recognition with oblique observation directions based on biomimetic pattern recognition

In this paper, we propose a new scheme for omnidirectional object-recognition in free space. The proposed scheme divides above problem into several onmidirectional object-recognition with different depression angles. An onmidirectional object-recognition system with oblique observation directions based on a new recognition theory-Biomimetic Pattern Recognition (BPR) is discussed in detail. Based on it, we can get the size of training samples in the onmidirectional object-recognition system in free space. Omnidirection ally cognitive tests were done on various kinds of animal models of rather similar shapes. For the total 8400 tests, the correct recognition rate is 99.89%. The rejection rate is 0.11% and on the condition of zero error rates. Experimental results are presented to show that the proposed approach outperforms three types of SVMs with either a three degree polynomial kernel or a radial basis function kernel.

Linux内核的实时支持的研究与实现

目前嵌入式计算的发展对实时多任务操作系统（RTOS）的需求日益迫切，而Linux内核的实时支持研究为迎合这种需求提供了新的可能性，通过从中断控制、实时调度策略、内核抢占机制及细粒度定时器几方面深入研究与实现了Linux内核的实时支持，以使之成为一个较为完善的RTOS。通过把实时部分设计为可配置的单元，以保留Linux原有系统支持及API。另外，为扩展系统适用范围，提出了可支持多实时调度策略的二阶段调度方案，同时指出了当前方案的不足和今后的研究方向。

基于核方法的敏感信息过滤的研究

在深入分析敏感信息过滤任务的特点和难点的基础上,针对现有一般的信息过滤方法的不足,提出了一种利用敏感词的组合信息来改进过滤效果的思想.进而,研究了在核方法的框架下特征共现行为建模的原则并提出了复合ANOVA核来刻画特征组合行为.通过真实信息过滤环境中的测试评估,显示了此敏感信息过滤方法的有效性.

面向入侵的取证系统框架

在分析常见入侵攻击的基础上抽象出入侵过程的一般模式，提出针对入侵攻击的取证系统应满足的特征。提出了入侵取证模型，并基于这一取证模型在操作系统内核层实现了取证系统原型KIFS（kernel intrusion forensic system）。在对实际入侵的取证实验中，根据KIFS得到的证据，成功记录并重构了一个针对FreeBSD系统漏洞的本地提升权限攻击的完整过程。

近邻边界Fisher判别分析

将数据集进行合理的维数约简对于一些机器学习算法效率的提高起着至关重要的影响.该文提出了一种利用数据点邻域信息的线性监督降维算法:近邻边界Fisher判别分析(Neighborhood Margin Fisher Discriminant Analysis,NMFDA).NMFDA尝试将每一数据点邻域内最远的同类数据点和最近的异类数据点之间的边界在投影子空间内尽可能地扩大,从而提高基于距离的识别算法的准确率.同时为了解决非线性降维问题,提出了Kernel NMFDA,通过在几个标准人脸数据库上与其它降维算法的对比识别实验,验证了提出算法的有效性.

旅游相关活动对九寨沟核心景区植物多样性与结构的影响

日益增强的旅游活动干扰正成为九寨沟世界自然遗产的有效保护与持续管理不可回避的挑战，已成为当前区域生态保护与经济发展的焦点问题之一。阐明相关旅游的干扰活动对核心景区植物物种组成、多样性与结构的影响是九寨沟有效保护与持续管理的必要前提和基础，然而至今少有研究。选择九寨沟与旅游活动相关的九个景点、公路建设地段以及退耕地还林地，详细调查了林下植被结构、物种组成与多样性，比较了相关旅游活动干扰与基本未受干扰地段的差异。目的是阐明九寨沟核心景区旅游干扰条件下植物多样性与群落结构及其特点，揭示旅游干扰与植被结构和生物多样的相互关系，评估九寨沟旅游管理的有效性，探索减免旅游干扰影响的对策与措施。初步结论如下。 1.旅游活动引起九寨沟核心景区植物组成和群落结构特征改变显著。栈道及公路附近许多耐荫喜湿的敏感种局部消失，而早熟禾、车前草、委陵菜等耐干旱、耐践踏、繁殖能力强的植物种群扩大；外来物种频繁出现并已少量侵入干扰相对较轻的林分深处；林下群落以草本植物为优势，灌木与苔藓植物的频度、盖度、高度以及灌木密度均有所下降。可见，大量的游人活动和景区公路建设已对九寨沟丰富的乡土植物构成极大威胁，导致其生物多样性降低。而九寨沟农耕地退耕有利于九寨沟生态环境和生物多样性保护，正逐步向稳定群落演替。 2．干扰强度的差异导致九寨沟植物所受影响的程度不同。栈道附近植物受影响的程度与游人活动频率有关，活动频率高的地段干扰强度大，对植物的影响程度重，反之植物所受的影响较轻。强度干扰地段，耐荫喜湿的物种少见，伴生植物优势地位突出，生物多样性明显降低，植物群落结构特征改变也极为显著；干扰较轻的地段，伴生物种少量出现，植物群落结构变化不明显，生物多样性略为降低，部分地区结构参数值和多样性指数有所升高。公路修建是一种强度干扰，它导致附近植物种类极为单一，草本优势种异常突出，多数植物生活力低下、生长更新能力差。 3.不同植物类群受干扰影响的程度不同。草本及苔藓植物的种类组成和多样性指数受干扰影响较大，灌木和苔藓植物的结构受干扰影响较大，苔藓植物对干扰影响最为敏感。 综合分析表明，九寨沟核心景区的管理虽然比较规范，但目前核心区热点景点段的管理仍然不够，旅游活动驱动了林下植被退化明显、物种组成显著变化、生物多样性衰退、非乡土喜光耐旱种群扩大。导致九寨沟核心区旅游活动与生物多样性保护目标尖锐冲突，进一步约束旅游活动带来的干扰，强化管理，开展林下植被恢复与非乡土喜光耐旱种群调控是九寨沟自然遗产地保护一项紧迫任务。 The increasing tourism disturbance is an unavoidable challenge to effective conservation and sustained tourist management of Jiuzhaigou Nature Reserve. It has become one of the focal problems of regional ecological protect and economic development. It is important to clarify effects of tourism disturbance on plant species composition, diversity and community structure in kernel spot for effective conservation and sustained tourist management in Jiuzhaigou, China. However, there were little studies about this yet. The study investigated the vegetation structure, species composition and diversity at nine sight spot, road area and four abandoned farmlands connecting with tourism, and compared the differences between disturbed area and undisturbed area. The purpose of the this study is clarifying the plant diversity and community structure and characteristics of the disturbed area in kernel spot of Jiuzhaigou, discovering the relation between vegetation structure and biodiversity, evaluating the effect of tourist management and exploring the measure decreasing tourist disturbance. Our results are following: 1. Tourism disturbance caused a significant change in species composition and structure of plant communities in kernel Spot of Jiuzhaigou. In the vicinity of plank and road, some native shade-tolerant or hygrophilous plants had disappeared, accompanying with the population expansion of some xerophilous and disturbance-resisting species such as Poa sp. Plantago depressa, Potentiila multicaulis and some exotic and synanthropic species. Herbs were domaint species, while frequency coverage and height of shrubs and bryophytes, and shrub density decreased. In indicated that tourist activities and build of road had adversely affected on native plant species, and led to decline in biodiversity of Jiuzhaigou Nature Reserve. Abandoned farmlands maybe conduced to entironment and biodiversity conservation. 2. Nearby the plank, influency variable of tourism disturbance on plant was alosely and positively correlated with disturbance intensity. There was companion plant species and lacked shape-loving species in heavy disturbed areas, which caused decrease in biodiversity and significant change in community structure in these places. On the contrary, in the slightly disturbed areas, some companion speices displayed and biodiversity decreased slightly, but no significant change in community structure in these areas. The biuld of road is a heavy disturbance form, which led to increase of herb species and to decrease in vitality and regeneration capacity. 3. The intensity of tourism disturbance on plant depended on plant groups. Tourism interference significantly influenced species composition and diversity index of herb and bryophyte; it also significantly influenced community structure; the bryophyte was more sensitive to tourism disturbance. Our result indicated that the management in kernel spot of Jiuzhaigou is relatively reasonable, but not adequate. Tourism speed the degradation of the vegetation under woodland, the change of the species composition, the decreaing of the biodiversity and the expanding of the exotic sunloving plant populations. The contradiction between tourism and conserving biodiversity is increasing, so enforcing management, regenerating the vegetation under woodland and adjusting the exotic sunloving arid-tolerent plant populations is a pressing work to protect the Jiuzhaigou natural legacy.

水稻同源四倍体三系法杂种优势利用研究

水稻是重要的粮食作物，其产量的增加和品质的改良都是关系国计民生的大事。就我国现阶段的国情而言，水稻产量在现有水平上稳步提升仍是未来十几年甚至几十年农业生产最重要的目标之一。尽管根据“超级杂交水稻育种”的战略设想和水稻育种实践，通过不断地改进育种技术可望在更高的产量水平上进行水稻杂种优势利用，在稻属植物内还具有很大的产量潜力可以挖掘。然而，仅仅从现有的种质基础出发，要更大幅度提高水稻单产，实现“超级杂交稻”的目标也存在一些困难：现有的推广品种是二倍体，尽管种类众多，但是其基因组的来源相对单一；同时，水稻基因组DNA含量也是作物中最少的，基因组内寻求开发潜力有一定困难；水稻作为C3植物，光合利用效率不高也是制约水稻产量提高的因素之一。因此，寻求常规手段以外的技术突破或者方法创新，是实现“超级杂交稻”的目标的迫切需求。本研究利用秋水仙素能抑制细胞分裂中纺锤丝的收缩、使细胞染色体加倍的作用，对水稻幼穗诱导的愈伤组织细胞进行加倍，并分化出再生植株；创制出水稻同源四倍体新的种质材料，在此基础上选育水稻同源四倍体雄性不育三系材料，并实现水稻同源四倍体的三系配套，开展水稻同源四倍体杂种优势利用和四倍体杂交水稻选育研究，建立水稻同源四倍体杂种优势利用的新技术体系。这不仅有助于倍性水平杂种优势的开拓和利用，同时也将为我国新世纪“超级稻”育种研究开辟一条新的技术途径。 水稻幼穗诱导愈伤组织并分化成苗是一项成熟、简单的组织培养技术。本研究以普通二倍体水稻亲本为材料，用秋水仙素进行水稻的多倍体化诱导，创制同源四倍体水稻三系亲本材料并对其进行鉴定。多倍体化以秋水仙素诱导的愈伤组织培养为基础，研究不同秋水仙素浓度梯度和愈伤组织诱导培养基组合对诱导四倍体植株的影响。结果表明在MS+2,4 D 1.0mg/L+ KT0.2mg/L+ IAA0.2mg/L 和500mg/L的秋水仙素处理下，水稻愈伤组织染色体加倍（有最高的效率）效果较好，平均加倍频率可达25.26%，其中，材料CDR22和IR26诱导较易成功，加倍频率分别达到75%和26.5%；相对材料94109 1.3%加倍频率和冈46B 10.8%加倍频率，诱导率差异极显著。 对水稻四倍体材料进行了形态学鉴定结果表明，与二倍体水稻对照相比其株高、穗长、花粉育性等主要农艺性状，确定四倍体材料在穗长和千粒重两方面极显著提高，种子的长度和宽度也显著增长。对花粉育性鉴定，确认水稻四倍体不育系材料仍为不育，保持系材料自交和杂交可育，恢复系材料自交和杂交可育。对四倍体材料进行细胞形态、染色体数目等方面进行细胞学鉴定，经核型分析表明水稻四倍体材料具有48条染色体，是二倍体水稻的两倍。水稻四倍体材料根尖分生组织细胞与二倍体的根尖分生组织细胞相比，细胞体积、细胞核和核仁显著增大。四倍体三系材料在细胞有丝分裂中期均可规则排列在赤道板，并能均等地移向两极；后期观察中没有发现染色体分离滞后现象，分裂末期细胞能够形成大小相对均一的子细胞。水稻同源四倍体三系材料细胞分裂未见异常，植株生长发育正常。 从1996年至2006年，针对结实率、有效分蘖、着粒数和穗长等主要农艺性状，通过系谱选育的方法，对培育的同源四倍体水稻亲本材料进行了连续选择和改良，取得较好成效。表现为结实率的改良效果极佳，所有改良材料的平均结实率均呈上升趋势，如D237（29.70％→72.70％）、DTB（19.55％→53.21％）等。有效分蘖总体呈现上升趋势，但在不同的年份，如1998和2002存在较大的负向波动。部分材料改良效果明显，如D19B（5.87→13.50）、D什香 (7.00→12.00)等；同时一些材料如DTB和D明恢63虽然总体略有提高，但在不同的年份波动很大，因此存在较大改良阻力，原因还有待进一步研究。着粒数的改良上升趋势比较显著，除保持系的DTB之外，其余材料的平均着粒数有显著提高。穗长的改良阻力较大，虽然不同材料总体上有所提高，但效果并不显著，并且不同年份有较大负向波动（2001）。此外还对株高、剑叶长等性状也进行了选择，但效果不显著，原因有待进一步提高。同源四倍体材料产量相关性状遗传改良幅度不一致，保持系和恢复系间的遗传改良效果也存在差异。这为同源四倍体水稻的进一步利用打下了良好的基础。 籼稻和粳稻亚种间杂交及杂种优势利用的主要障碍就是其低的结实率。而同源四倍体杂交水稻的研究为提高杂交水稻的杂种优势利用创造了新的途径。本研究通过随机区组设计方案，挑选性状优良的二倍体水稻材料，包括雄性不育系，保持系和恢复系进行秋水仙素诱导加倍，从而获得同源四倍体水稻对应的三系材料。利用选育的优良水稻同源四倍体三系材料，配制7个杂交组合，杂交F1代与其恢复系亲本进行比较，用于计算超亲优势（HB）；而杂交F1代与生产上大面积推广的二倍体杂交品种汕优63进行比较，用于计算杂种优势。结果显示，同源四倍体杂交水稻的超亲优势表现为：每株有效穗变化幅度为1.4%至105.9%，总粒数为0.5%至74.3%，每穗实粒数为17.6%至255.7%，结实率为9.6%至130.4%。这些农艺性状的改良使得这7个杂种F1的理论产量的超亲优势高达64.8%至672.7%。小区试验中四倍体杂交水稻组合T461A/T4002和T461A/T4193分别比二倍体对照汕优63提高46.3%和38.3%以上，除一个品种以外所有品种产量均接近或高于汕优63的产量。同源四倍体水稻强大的杂种优势表明，亚种间杂交育性低的问题可通过四倍体化及强化选择来解决。此外，同源四倍体杂交水稻器官的巨大性也是其产量提高的有利因素，水稻同源四倍体三系杂种优势利用研究具有一定的理论价值和商业生产潜力。 Rice is one of the major food crops, the improvement of the production and quality of it is an important thing related to the people's livelihood. On China's current national conditions, steadily increase of the rice yield based on the current level is still one of the most important goals in the next decade or even decades of agricultural production. According to the "super hybrid rice breeding" the strategic and rice breeding practice, improvement of the use of hybrid rice heterosis through continuous improvements in breeding technology is expected to get a higher level of rice yield, there are also a great yield potential can be exploited. However, there are also some difficulties to increase rice yield obviously and implement the goal of "super hybrid rice" based on the existing germplasm: Rice varieties in promotion are diploid, although there are many varieties, but their genome are from a comparatively single source; Meanwhile, the rice genome DNA are the least among the crops, it is difficult to exploit the development potential within the genome; Rice as C3 plants, photosynthetic efficiency is not high, it is one of the factors constraint rice yield. Therefore, seeking technological breakthroughs or innovative methods different from conventional means is the urgent needs to reach the target of "super hybrid rice". Using colchicine inhibit spindle contraction during cell division, double the cell chromosome, we induced callus cells from rice panicle to be doubled, and differentiated regeneration; we created a new autotetraploid rice germplasm material, and on that basis we bred male sterility three line autotetraploid rice materials, and the achieved the three line rice autotetraploid matchmaking, researched in autotetraploid rice heterosis usage and tetraploid hybrid rice breeding, constituted a new technology system of autotetraploid hybrid rice heterosis utilization. This not only helps the tetraploid rice heterosis exploration and use, but also inaugurates a new technical means for China in the new century "super rice" breeding research. We chose ordinary diploid rice as materials, using colchicine to induce the polyploidization, created the autotetraploid rice three-line materials and identified them. The polyploidization was based on the colchicine-induced callus tissue culture, and we experimented different colchicine concentrations and culture mediums to induce tetraploid plants, confirmed that the optimal concentration for inducement was 500 mg/L, the average induce rate was 25.26 %. Among all the materials, CDR22 and IR26 had higher induced rate; in contrary, 94109 and GANG46B had lower induced rate, the difference was significant. Autotetraploid materials was identified of both morphological and cytological, compared plant height, length of pollen sterility, and other major agronomic traits with a diploid rice as the control plant, identified that the autotetraploid materials had very significant advantages in ear length and thousand-grain weight, as well as the size of the seeds. Cytology identification included observation of the cell morphology, the number of chromosomes, and karyotype analysis on the autotetraploid materials confirmed that their chromosome number was 48, twice of the diploid rice. Mitoses in the three lines were common: chromosomes arrayed normally in metaphase and separated balanced into the two poles, chromosome moved without lagging in anaphase and daughter cells normally formed in telophase except one. It has been proved that tetraploid rice has normal meiosis as their diploid relatives, which usually including series of sub-phases as interphase, prophase I (five sub-phases), prophase II, metaphase I, II, anaphase I, II and telophase I, II. However, abnormal phenomena, such as formation of tetravalent, trivalent and univalent, chromosome lagging and so on, which would finally block meiosis. Configurations of chromosome in metaphaseⅠwere versatile in structure and form accept the bivalent. That condition varied in different strain, suggesting more complex paring configurations and more versatile genetic characters in tetraploid rice. All these abnormalities in meiosis contributed to low fertility of gamete and might consequently resulted in low seed setting. Successive selection and improvement on seed set, productive tiller per plant, total grains per panicle, panicle length and so on had been carried out from 1996 to 2006. The raise of seed sets was significant in both restorers and maintainers. Seed sets of some strains were improved more significantly than others, for example D237（29.70％→72.70％）、DTB（19.55％→53.21％）and et al.. Productive tiller per plant was improved to some extant. The tendency of improvement was rising on the whole but changed in some years such as 1998 and 2002. Part of the stains increased greatly, such as D19B（5.87→13.50）、Dshixiang (7.00→12.00) and so on, but some strains including DTB and Dminghui63 only increased little and decreased in some years by unknown reason. Total grains per panicle increased significantly and all strains except DTB increased. Improvement of panicle length termed to be hard. Different strains showed different capacities for improvement and floating existed in different years for example 2001. It has been proved that other agronomical traits including plant length, flag leaf length and so on could be improved but not significantly by selection also. In a word, agronomical traits could be raised by successive selection that is prerequisite for further utility of autotetraploid rice. Poor fertility is the main barrier for utilizing heterosis between the two rice (Oryza stiva L.) sub-species, indica and japonica. Recently, the development of autotetraploid hybrids (2n=4x=48) has been suggested as a new method for increasing heterosis in hybrid rice. Using standard experimental protocols, the elite diploid rice male sterile, maintainer, and restorer lines were colchine-doubled and autotetraploid counterparts were obtained. Seven resulting hybrids were analyzed for heterobeltiosis (HB), where the F1 was compared to the male parent, and the degree of heterosis, where the F1 was compared to the diploid commercial hybrid, Shanyou 63. The HB among the autotetraploid hybrids ranged from 1.4 to 105.9% for the productive panicles per plant, 0.5 to 74.3% for total kernels per panicle, 17.6 to 255.7% for filled kernels per panicle, and 9.6 to 130.4% for seed set. Improvements in these yield components resulted in the HB for kernel yield ranging from 64.8 to 672.7% among the seven hybrids. Hybrids T461A/T4002 and T461A/T4193 yielded 46.3 and 38.3% more, respectively than Shanyou 63, and all other hybrids but one yielded the same or more than Shanyou 63. The high heterosis for yield suggests that hybrid sterility between two rice sub-species may be overcome by using tetraploid lines followed by intensive selection. Also, the gigantic features of the autotetraploid hybrids may establish a plant structure able to support the higher yield.