Two-dimensional network of dislocations and nanocavities in hydrogen-implanted and two-step annealed silicon

Conventional transmission electron microscopy and energy-filtering were used to study the dislocations and nanocavities in proton-implanted [001] silicon. A two-dimensional network of dislocations and nanocavities was found after a two-step annealing, while only isolated cavities were present in single-step annealed Si. In addition, two-step annealing increased materially the size and density of the nanocavities. The Burgers vector of the dislocations was mainly the 1/2[110] type. The gettering of oxygen at the nanocavities was demonstrated. (C) 1998 American Institute of Physics. [S0003-6951(98)00620-2].

A Novel CMOS Color Pixel for Vision Chips

This paper presents a novel CMOS color pixel with a 2D metal-grating structure for real-time vision chips. It consists of an N-well/P-substrate diode without salicide and 2D metal-grating layers on the diode. The periods of the 2D metal structure are controlled to realize color filtering. We implemented sixteen kinds of the pixels with the different metal-grating structures in a standard 0.18 mu m CMOS process. The measured results demonstrate that the N-well/P-substrate diode without salicide and with the 2D metal-grating structures can serve as the high speed RGB color active pixel sensor for real-time vision chips well.

方差阴影图中的光渗现象消除算法

方差阴影图算法使用概率的方法计算像素被遮挡的上限概率,通过对深度图滤波的方法来有效地减少阴影图算法中的走样问题,但在深度比较复杂的场景中方差阴影图算法会出现光渗现象,即在应该是阴影的区域却有了亮度.文中使用最小-最大阴影图来辅助消除方差阴影图中的光渗现象,在对深度纹理进行滤波的同时生成一个最小-最大阴影图;在实时绘制场景时,利用最小-最大阴影图来辅助判断当前片元是否完全处在阴影区域内部,由此生成更真实、更准确的阴影.该算法可以很容易地添加到已有的方差阴影图算法的片元处理程序中,并且不会对原有阴影的柔和边界以及绘制的帧率产生影响.

基于CoP建模的协作过滤推荐方法

传统的协作过滤推荐方法主要基于个人兴趣特征来实现推荐。在组织内部协作场景下,为实现知识共享与重用,推荐系统不仅要考虑用户兴趣,还应考虑用户和用户组的任务。传统的协作过滤推荐方法已不能满足要求。CoP是组织内部人员管理的主要形式,它的特征是其成员任务的反映。基于已有的协作过滤推荐研究与D-S理论,提出了一种CoP特征构建算法,并以此为基础研究了面向CoP的协作过滤推荐。

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.

增量式个性化推荐系统研究

互联网个性化推荐系统(Internet personal recommender systems)是根据用户的兴趣推荐最相关的互联网信息给用户的系统。在网上信息过载矛盾越来越严重、用户信息检索的个性化需求日益增强的现状下，推荐系统已经在搜索引擎、电子商务、网上社区等互联网关键应用中起到了关键性的作用，并且越来越受到重视。 然而，在大型网站上部署一个成熟推荐系统的代价依然很大，需要大量的计算和存储资源，推荐的准确性也依然有很大提升空间和需求，这就为推荐系统的研究提供了很多挑战。在这些挑战中推荐算法的准确性和可扩展性一直是该领域最为关注的两个问题，所谓推荐的准确性是指推荐的信息中用户真正感兴趣的比例，而可扩展性指的是系统能否在可容忍的时间和空间复杂度内处理海量的数据。如何在提高算法推荐准确性的同时增强算法的可扩展性是推荐系统改进的主要研究目标。然而，目前学术界的研究更多侧重于提高推荐算法的准确性，而对于可扩展性，很多准确性很高的算法由于需要比较复杂的计算，处理大规模动态数据的能力往往比较有限，并且它们的评测实验中并没有将可扩展性纳入到评价范畴，导致这些算法目前还很难在工业界大规模应用。 本论文的研究试图解决这一问题。通过在推荐算法中借鉴增量学习(Incremental learning)的思想，即考虑最新的训练数据来更新原有的机器学习模型，不需要或仅需要参考部分旧的训练数据，相对于使用全部数据也即批量的处理方式，增量式改进可以大大降低模型更新的复杂度，从而可以大幅度提高推荐算法在遇到新的训练数据时推荐模型更新的效率，降低计算代价，使得推荐模型的更新可以更加及时，进而提高推荐结果的准确性。具体来说，我们在提出了两种新的增量式协同过滤算法的同时，采用增量式学习的方法对目前准确性最好的若干推荐算法进行加速，特别是提高这些算法面对新的训练数据的更新模型的速度和效率，从而为这些算法的大规模的应用提供了可能。另一方面，新的训练数据包含了最新的用户兴趣，因此相对于旧的训练数据，算法在做更新时应给予更高的权重，这样才能做到推荐的结果在考虑到用户长期兴趣的同时，特别考虑用户近期的兴趣，从而使得推荐结果更加准确。这两方面归纳起来，我们旨在通过增量式学习使得推荐算法在更新时更加高效和精确，真正适用于互联网上海量数据的推荐，同时对其他增量式推荐系统方面的研究也具有借鉴意义。我们的改进工作主要包括以下几个方面： 基于主题模型的增量式推荐算法。主题模型，特别是概率隐含主题模型(PLSA)是一种广泛应用于推荐系统的主流方法，在文本推荐、图像推荐以及协同过滤推荐领域都有着很好的推荐效果。目前制约PLSA算法取得更大成功的重要因素就是PLSA算法更新的复杂度过高，使得学习模型的更新只能做批量式处理，这样就导致推荐的时效性不高，也没有办法体现用户的最新的兴趣和整体的最新动态。我们提出了一种增量式学习方法，可以应用于文本分析领域和协同过滤领域，当有新的训练数据到来时，对于基于文本的推荐，增量式更新方法仅寻找最相关的用户和文本以及涉及到的单词进行主题分布的更新，并给予新的文本以更高权重；对于协同过滤，我们的方法仅对当前用户所评分过得物品以及当前物品所涉及的用户进行更新，大大降低了更新的运算复杂度，提高了新数据在推荐算法中所占的权重，使得推荐更加准确、及时。我们的算法在天涯问答文本数据集上和MovieLens电影推荐数据集、Last.FM歌曲推荐数据集、豆瓣图书推荐数据集等协同过滤数据集上取得了很好的效果。 基于蚁群算法(Ant colony algorithm)的协同过滤推荐方法。受到群体智能(Swarm intelligence)算法的启发，我们提出了一种类似于蚁群算法的协同过滤推荐方法——Ant Collaborative Filtering，初始化阶段该方法给予每个用户或一组用户以全局唯一的单位数量的信息素，当用户对物品评分或者用户表示对该物品感兴趣时，用户所携带的信息素相应的传播到该物品上，同时该物品上已有的信息素（初始化为0）也会相应的传播给该用户；此外，用户和物品所携带的信息素会随着时间的推移有一定速率的挥发，通过挥发机制，可以在推荐时更重视用户近期的兴趣；推荐阶段，按照用户和物品所携带的信息素的种类和数量，我们可以得到相应的相似度，进而通过经典的相似度比较的方法来进行推荐。基于蚁群的协同过滤方法的优势在于可以有效的降低训练数据中的稀疏性，并且推荐算法可以实时的进行更新和推荐，同时考虑了用户兴趣随着时间的变化。我们在MovieLens电影评分、豆瓣书籍推荐、Last.FM音乐推荐数据集上验证了我们的方法。最后，我们建立了一个互联网新闻推荐系统，该系统以Firefox插件形式实现，自动采集用户浏览兴趣和偏好，后端使用不同的推荐算法推荐用户感兴趣的新闻给用户。 基于联合聚类(Co-clustering)的两阶段协同过滤方法。聚类(Clustering)是一种缩小数据规模、降低数据稀疏性的有效方法。对于庞大而稀疏的协同过滤训练数据来说，聚类是一种很自然事实上也的确很有效的预处理方法。因此我们提出了一种两阶段协同过滤框架：首先通过我们提出的一种联合聚类的方法，将原始评分矩阵分解成很多维度很小的块，每一块里面包含相似的用户对相似的物品的评分，然后通过矩阵拟合的方法（我们使用了非负矩阵分解NMF和主题模型PLSA）来对这些小块中的未知评分进行预测。当用户新增了对于某物品的一条评分，我们仅需要更新该用户或该物品所处的数据块进行重新评分预估，大大加快了评分预估的速度。我们在MovieLens电影评分数据集上验证了该算法的效果。 本文的研究成果不仅可以直接应用于大型推荐系统中，而且对于增量式推荐系统的后续研究也具有一定的指导意义。首先基于PLSA的增量式推荐算法对于其他基于图模型的推荐系统具有借鉴价值，其次蚁群推荐算法为一类新的、基于群体智能(Swarm intellignece)的协同过滤算法做出了有价值的探索，最后我们提出的两阶段协同过滤框架对于提高推荐算法的可扩展性和更新效率提出了一个通用的有效解决方案。 推荐系统是一个无止尽的优化的过程，除了推荐精度的不断提高之外，推荐算法的性能随着互联网上数据量的增加也需要进一步提高，增量式学习无疑是提高推荐算法更新速度最重要的方法，本文的研究为这一方向提供了参考。

Optical image processing using the photoinduced anisotropy of pyrrylfulgide

A synthesized photochromic compound-pyrrylfulgide-is prepared as a thin film doped in a polymethylmethacrylate (PMMA) matrix. Under irradiation by UV light, the film converts from the bleached state into a colored state that has a maximum absorption at 635 nm and is thermally stable at room temperature. When the colored state is irradiated by a linearly polarized 650 nm laser, the film returns to the bleached state; photoinduced anisotropy is produced during this process. Application of optical image processing methods using the photoinduced anisotropy of the pyrrylfulgide/PMMA film is described. Examples in non-Fourier optical image processing, such as contrast reversal and image subtraction and summation, as well as in Fourier optical image processing, such as low-pass filtering and edge enhancement, are presented. (c) 2006 Optical Society of America.

SOA-based polarity-preserving all-optical wavelength conversion at 80Gbit/s: Wide conversion range, well dynamic characteristics and polarization insensitive

This letter presents the effective design of a tunable 80 Gbit/s wavelength converter with a simple configuration consisting of a single semiconductor optical amplifier (SOA) and an optical bandpass filter (OBPF). Based on both cross-gain and cross-phase modulation in SOA, the polarity-preserved, ultrafast wavelength conversion is achieved by appropriately filtering the blue-chirped spectral component of a probe light. Moreover, the experiments are carried out to investigate into the wavelength tunability and the maximum tuning range of the designed wavelength converter. Our results show that a wide wavelength conversion range of nearly 35 nm is achieved with 21-nm downconversion and 14-nm upconversion, which is substantially limited by the operation wavelength ranges of a tunable OBPF and a tunable continuous-wave laser in our experiment. We also exploited the dynamics characteristics of the wavelength converter with variable input powers and different injection current of SOA. (C) 2008 Wiley Periodicals, Inc.

基于滤波方法的DEM尺度变换方法研究

针对水土流失评价中地形因子的尺度效应问题,引入数字图像处理中的滤波分析方法,对5 m分辨率DEM进行了滤波处理,并通过地形综合特征分析,坡度、曲率制图统计分析,派生等高线与河流提取统计分析等方法,对滤波处理的综合效果进行了分析论证。结果表明,利用滤波方法可以实现对高分辨率DEM的综合,利用比例尺因子为5或者10的处理,其结果与10 m分辨率DEM或1∶5万地形图在反映地形起伏方面相似。

CONSTRUCTION AND APPLICATIONS OF A DUAL MASS-SELECTED LOW-ENERGY ION-BEAM SYSTEM

A direct ion beam deposition system designed for heteroepitaxy at a low substrate temperature and for the growth of metastable compounds has been constructed and tested. The system consists of two mass-resolved low-energy ion beams which merge at the target with an incident energy range 50-25 000 eV. Each ion beam uses a Freeman ion source for ion production and a magnetic sector for mass filtering. While a magnetic quadrupole lens is used in one beam for ion optics, an electrostatic quadrupole lens focuses the other beam. Both focusing approaches provide a current density more than 100-mu-A/cm2, although the magnetic quadrupole gives a better performance for ion energies below 200 eV. The typical current of each beam reaches more than 0.3 mA at 100 eV, with a ribbon beam of about 0.3-0.5 x 2 cm2. The target is housed in an ultrahigh vacuum chamber with a base pressure of 1 x 10(-7) Pa and a typical pressure of 5 x 10(-6) Pa when a noncondensable beam like argon is brought into the chamber. During deposition, the target can be heated to 800-degrees-C and scanned mechanically with an electronic scanning control unit. The dual beam system has been used to grow GaN using a Ga+ and a N+ beam, and to study the oxygen and hydrogen ion beam bombardment effects during carbon ion beam deposition. The results showed that the simultaneous arrival of two beams at the target is particularly useful in compound formation and in elucidation of growth mechanisms.

Structural and infrared absorption properties of self-organized InGaAs GaAs quantum dots multilayers

Self-organized InGaAs/GaAs quantum dots (QDs) stacked multilayers have been prepared by solid source molecular beam epitaxy. Cross-sectional transmission electron microscopy shows that the InGaAs QDs are nearly perfectly vertically aligned in the growth direction [100]. The filtering effect on the QDs distribution is found to be the dominant mechanism leading to vertical alignment and a highly uniform size distribution. Moreover, we observe a distinct infrared absorption from the sample in the range of 8.6-10.7 mu m. This indicates the potential of QDs multilayer structure for use as infrared photodetector.

High sensitivity beam phase measurement system for isochronous cyclotron

The beam phase measurement system in the HIRFL is introduced. Based on the double-balanced mixer principle using rf-signal mixing and filtering techniques, a stable and sensitive phase measurement system has been developed. The phase history of the ion beam is detected by using a set of capacitive pick-up probes installed in the cyclotron. The phase information of the measurement is necessary for tuning to obtain a optimized isochronous magnetic field which induces to maximize the beam intensity and to optimize the beam quality. The result of the phase measurement is reliable and the accurancy reaches +/- 0.5 degrees.

2-D and 3-D processing and interpretation of multi-fold ground penetrating radar data:a case history from an archaeological site

A 2.5-D and 3-D multi-fold GPR survey was carried out in the Archaeological Park of Aquileia (northern Italy). The primary objective of the study was the identification of targets of potential archaeological interest in an area designated by local archaeological authorities. The second geophysical objective was to test 2-D and 3-D multi-fold methods and to study localised targets of unknown shape and dimensions in hostile soil conditions. Several portions of the acquisition grid were processed in common offset (CO), common shot (CSG) and common mid point (CMP) geometry. An 8×8 m area was studied with orthogonal CMPs thus achieving a 3-D subsurface coverage with azimuthal range limited to two normal components. Coherent noise components were identified in the pre-stack domain and removed by means of FK filtering of CMP records. Stack velocities were obtained from conventional velocity analysis and azimuthal velocity analysis of 3-D pre-stack gathers. Two major discontinuities were identified in the area of study. The deeper one most probably coincides with the paleosol at the base of the layer associated with activities of man in the area in the last 2500 years. This interpretation is in agreement with the results obtained from nearby cores and excavations. The shallow discontinuity is observed in a part of the investigated area and it shows local interruptions with a linear distribution on the grid. Such interruptions may correspond to buried targets of archaeological interest. The prominent enhancement of the subsurface images obtained by means of multi-fold techniques, compared with the relatively poor quality of the conventional single-fold georadar sections, indicates that multi-fold methods are well suited for the application to high resolution studies in archaeology.

基于I-Match算法的垃圾邮件过滤研究

电子邮件(Electronic Mail，E-Mail)是目前使用最广泛的互联网应用。随着互联网络以惊人的速度增长，电子邮件成为发布恶意信息的一个重要途径，垃圾邮件已经成为危害互联网络的最大毒瘤。针对方式多样的垃圾邮件技术，垃圾邮件过滤系统往往也需要综合多种过滤技术以提高系统的有效性。其中摘要技术已经成为重要的垃圾邮件过滤方法之一：通过摘要技术判断一个邮件和已知垃圾邮件的相似度，从而对邮件进行分类。判断一个垃圾邮件过滤算法是否有效，要综合考虑算法的召回率、准确率以及时间性能。I-Match算法通过摘要值的精确匹配来判断两个邮件文本内容是否相似，算法在效率方面表现突出。但是I-Match算法在实际的应用中还存在很多问题，其中包括字典生成制约算法的性能以及面对攻击时算法表现出的鲁棒性不足。因此，优化算法的字典生成过程以及提高算法的鲁棒性成了算法应用于实际系统的两个重要问题。本文的主要工作包含以下内容： 对垃圾邮件进行相似性分析，包括垃圾邮件相似性的起因、垃圾邮件在时间和内容两方面所表现出的相似性特征。垃圾邮件体现出的相似性特征是使用摘要算法进行垃圾邮件过滤的必要条件之一。 改进I-Match算法的字典生成过程。提出利用特征的互信息作为特征选择依据改进字典生成过程，并对比几种不同的特征选择方式对算法性能的影响。 分析I-Match算法的鲁棒性以及几种I-Match改进算法对算法鲁棒性的提升，在实际的邮件语料上对各种改进算法进行评测，并综合分析各个算法的实用性。 完成了KSpam系统原型，以插件的形式综合多种邮件过滤方法，并给出了I-Match算法在KSpam系统中的实现方案。同时，系统实现了一种新式的邮件自动回收功能，有效减少邮件管理员的邮件语料收集工作。