Relative density-based classification noise detection

Classification noise is a common byproduct of traditional data mining approaches, and no specialized approach for detecting classification noise is currently available. Methods for outlier detection are well-developed, but outliers and classification noise have characteristics different enough to make outlier detection algorithms unsuitable for classification noise detection. In this paper, a new, specialized approach to detect classification noise is proposed, named relative density based classification noise detection (RDBCND). Computational experiments in artificial data sets described herein show that RDBCND has time complexity of O(n log n), indicating greater efficiency than traditional approaches, which exhibit time complexity of at least O(n²). The use of classification noise detection to improve the generalization ability of common classifier algorithms is also described. In particular, a new unified approach based on RDBCND is compared to a cross validation approach applied to a BP neural network. Trials in both artificial and real-life datasets show that the RDBCND-based approach can greatly accelerate the process of identifying the best decision function. The novel method can also eliminate underfitting, as the algorithm simply searches for the highest training accuracy. The experiments also show that the RDBCND-based method has greater accuracy and lower cpu time in reaching global solutions than the cross-validation method. Since the relative density is a local concept, our new approach can be directly used in nonlinear datasets without data transformation. It is a great advantage compared to some linear classifier algorithms. As in current linear classifiers, the kernel functions or other transformations need to be used to make them suitable for non-linear datasets, and that will increase their complexity.     

A study on small-world brain functional networks altered by postherpetic neuralgia

Understanding the effect of postherpetic neuralgia (PHN) pain on brain activity is important for clinical strategies. This is the first study, to our knowledge, to relate PHN pain to small-world properties of brain functional networks. Functional magnetic resonance imaging (fMRI) was used to construct functional brain networks of the subjects during the resting state. Sixteen patients with PHN pain and 16 (8 males, 8 females for both groups) age-matched controls were studied. The PHN patients exhibited decreased local efficiency along with non-significant changes of global efficiency in comparison with the healthy controls. Moreover, regional nodal efficiency was found to be significantly affected by PHN pain in the areas related to sense (postcentral gyrus, inferior parietal gyrus and thalamus), memory/affective processes (parahippocampal gyrus) and emotional activities (putamen). Significant correlation (p < 0.05) was also found between the nodal efficiency of putamen and pain intensity in PHN patients. Our results suggest that PHN modulates the local efficiency, and the small-world properties of brain networks may have potentials to objectively evaluate pain information in clinic.     

红外双波段双层谐衍射光学系统设计

将谐衍射透镜应用在传统红外单波段佩茨瓦尔(Petzval)物镜上,设计得到工作波段处于3.4~4.2μm和8~11μm的红外双波段单层谐衍射光学系统。但单层谐衍射元件的衍射效率只在设计波长处衍射效率最高,随着波长相对设计中心波长向两侧偏离,主衍射级次的衍射效率逐渐下降。为提高含单层谐衍射元件光学系统的衍射效率,基于双层衍射元件衍射效率表达式研究了双层谐衍射元件的结构优化,给出了优化方法。设计出佩茨瓦尔型红外双波段双层谐衍射光学系统,其在3.4~4.2μm和8~11μm两个工作波段的衍射效率均达到90%以上,相比含有单层谐衍射面的光学系统衍射效率有了很大提升,提高了像面衬比度,完善了系统成像质量。     

1 053-nm all-fiber pulse multi-pass stretcher using a linear chirped fiber Bragg grating

An all-fiber optical laser pulse multi-pass stretcher using a chirped fiber Bragg grating (CFBG) is demonstrated. Pulses from a 1 053-nm mode-locked fiber seed oscillator are stretched by multiple passing through a chirped fiber grating set in a fiber regenerative amplifier structure. We stretch the pulse from 16 ps to 1.855 ns after it transmits seven loops in the stretcher. The main factors that affect the stretching results are discussed.     

超高精细度微光学腔共振频率及有效腔长的精密测量

超高精细度微共振器是实现原子或者其他偶极子与腔强耦合作用的基本部分, 在腔量子电动力学(QED)、弱光非线性效应及微光学器件研究中扮演着重要的角色. 微腔基本参数的精密测量最终可以确定腔与原子的耦合系数、腔场衰减率, 对决定系统的动力学特性具有重要的意义. 但是由于超高精细度光学微腔本身的构造和多层镀膜的特点, 高精度地确定其共振频率及有效腔长存在一定困难. 本文结合修正的多层介质膜模型, 实验上完成了膜层为37层的超高精细度光学微腔在不同共振频率下有效腔长的精密测量, 获得了超高精细度光学微腔的共振频率及波长; 理论计算分析与实验测量结果相符, 对纵模间隔的测量精度误差低于0.004 nm, 较为修正前提高了约两个量级. 同时给出了对应不同模式数下, 光波渗入到介质中的深度. 该方法可望应用到其他微共振器的精密测量中. 关键词： 光学微腔 高精细度 共振频率     

基于超图结构的科研合作网络演化模型

基于科研论文作者的合作方式, 用超图理论构建了一个科研合作超网络演化模型. 利用平均场理论分析了作者发表论文的演化规律, 发现作者的超度 (即发表论文数) 分布符合幂律分布. 进一步理论分析得到分布的幂指数γ与合作领域作者增长速度相关. γ越大, 新作者增长速度越快, 且存在关系: γ=1+L/M (L/M为作者增长率). 并通过对《物理学报》与《中国科学》2003–2012年期间作者发表论文进行了数据分析, 实证结果与理论分析及模拟结果能很好地符合. 本文对科研合作网络的理论和实证研究有一定的借鉴意义. 关键词： 复杂网络 超图 科研合作网络 演化模型     

Quantum phase transitions of fermionic atoms in an anisotropic triangular optical lattice

The effect of anisotropy caused by a confining potential on the properties of fermionic cold atoms in a triangular optical lattice is systematically investigated by using the dynamical cluster approximation combined with the continuous time quantum Monte–Carlo algorithm.The quantum phase diagrams which reflect the temperature–interaction relation and the competition between the anisotropic parameter and the interaction are presented with full consideration of the anisotropy of the system.Our results show that the system undergoes a transition from Fermi liquid to Mott insulator when the repulsive interaction reaches a critical value.The Kondo effect also can be observed in this system and the pseudogap is suppressed at low temperatures due to the Kondo effect.A feasible experiment protocol to observe these phenomena in an anisotropic triangular optical lattice with cold atoms is proposed,in which the hopping terms are closely related to the lattice confining potential and the atomic interaction can be adjusted via the Feshbach resonance.     

A microscopic pedestrian-simulation model and its application to intersecting flows

We develop a microscopic pedestrian-simulation model in which pedestrian positions are updated at discrete time steps. At each time step, each pedestrian probabilistically selects a direction of movement from a predetermined set according to a logit-type function that considers the dynamics of other pedestrians around, and then selects a step size that satisfies a certain distribution. We perform a number of field experiments on real intersecting pedestrian flows with four different angles. We then validate and calibrate the model using sample data on the deviation angles, step velocities, and velocity–density relations obtained from the experiments.     

基于飞秒激光写制波导的PPKTP晶体倍频实验研究

在PPKTP晶体上进行了飞秒激光刻写波导的实验研究,优化了c-切PPKTP晶体中写制二型波导的工艺参数.当波导长度为10mm,宽度为14.5μm,准相位匹配波长为1064nm时,实现了单模传输.利用该波导对调QNd:YAG激光进行了准相位匹配倍频实验,实现的二次谐波转换效率为39.8%.