EcoLens: visual analysis of ecological regions in urban contexts using traffic data

Journal of Visualization - The increasing availability of spatiotemporal data provides unprecedented opportunities for understanding the structure of an urban area in terms of people’s...     

Model averaging for linear models with responses missing at random

Annals of the Institute of Statistical Mathematics - In this paper, a model averaging approach is developed for the linear regression models with response missing at random. It is shown that the...     

Preface: Mechanics of soft materials and flexible structures

Acta Mechanica Sinica -     

基于电子病历的乳腺癌群组与治疗方案可视分析

乳腺癌是当前最常见的恶性肿瘤之一，其电子病历数据可用于挖掘隐含规律，对治疗与预后分析有重要意义。通过与乳腺科医生合作，选择合适的预测模型和可视化方法，搭建了一个基于电子病历的乳腺癌群组和治疗方案可视分析系统。首先，对具有高维属性的病人进行降维和聚类处理，形成病人群组，并采用南丁格尔图、词云和时间轴可视化方法，直观展示病人群组间特征的差异；然后，用支持向量机（support vector machine，SVM）模型预测治疗方案，用平行坐标、矩阵热力图和分类图分别展示属性相关性、训练后的特征权重和预测结果；最后，用真实案例验证了系统在群组分析、治疗方案及病人属性关联分析中的有效性，从而较好地帮助医生选择合适的治疗方案。     

Point interactions for 3D sub-Laplacians

In this paper we show that, for a sub-Laplacian Δ on a 3-dimensional manifold M, no point interaction centered at a point $q_0\in M$ exists. When M is complete w.r.t. the associated sub-Riemannian structure, this means that Δ acting on $C_0^{\infty }(M?\{q_0\})$ is essentially self-adjoint in $L^2(M)$. A particular example is the standard sub-Laplacian on the Heisenberg group. This is in stark contrast with what happens in a Riemannian manifold N, whose associated Laplace-Beltrami operator acting on $C_0^{\infty }(N?\{q_0\})$ is never essentially self-adjoint in $L^2(N)$, if $\dim ?N\le 3$. We then apply this result to the Schrödinger evolution of a thin molecule, i.e., with a vanishing moment of inertia, rotating around its center of mass.     

Parametric modeling of steady-state gradient coil vibration: Resonance dynamics under variations in cylinder geometry

Gradient coil (GC) vibration is the root cause of many problems in MRI adversely affecting scanner performance, image quality, and acoustic noise levels. A critical issue is that GC vibration will be significantly increased close to any GC mechanical resonances. It is well known that altering the dimensions of a GC fundamentally affects the mechanical resonances excited by the GC windings. The precise nature of the effects (i.e., how the resonances are affected) is however not well understood. The purpose of the present paper is to study how the mechanical resonances excited by closed whole-body Z-gradient coils are affected by variations in cylinder geometry. A mathematical Z-gradient coil vibration model recently developed and validated by the authors is used to theoretically study the resonance dynamics under variation(s) in cylinder: (i) length, (ii) mean radius, and (iii) radial thickness. The forced-vibration response to Lorentz-force excitation is in each case analyzed in terms of the frequency response of the GC cylinder's displacement. In cases (i) and (ii), the qualitative dynamics are simple: reducing the cylinder length and/or mean radius causes all mechanical resonances to shift to higher frequencies. In case (iii), the qualitative dynamics are much more complicated with different resonances shifting in different directions and additional dependencies on the cylinder length. The more detailed dynamics are intricate owing to the fact that resonances shift at comparatively different rates and this leads to several novel and theoretically interesting predicted effects. Knowledge of these effects advance our understanding of the basic mechanics of GC vibration and offer practically useful insights into how such vibration may be passively reduced.     

Quantized-states-based adaptive control against unknown slippage effects of uncertain mobile robots with input and state quantization

An adaptive tracking design strategy based on quantized state feedback is developed for uncertain nonholonomic mobile robots with unknown wheel slippage effects. All state variables and control torques are assumed to be quantized by the state and input quantizers, respectively, in a network control environment. Thus, the quantized state feedback information is only available for the tracking control design. An approximation-based adaptive controller using quantized states is recursively designed to ensure the robust adaptive tracking against unknown wheel slippage effects where the quantized-states-based adaptive mechanism is derived to compensate for unknown wheel slippage effects, system nonlinearities, and quantization errors. The boundedness of the quantization errors and estimated parameters in the closed-loop system is analyzed by presenting some theoretical lemmas. Based on these lemmas, we prove the uniform ultimate boundedness of closed-loop signals and the convergence of the trajectory tracking error in the presence of wheel slippage effects. Simulations verify the effectiveness of the resulting tracking scheme.     

基于联合神经网络学习的中文电力计量命名实体识别

随着电力计量业务的不断扩展，迫切需要由业务信息、技术知识、行业标准及其内在联系所组成的电力计量知识图谱，为电网的决策和发展提供更为全面有效的支持。命名实体识别是构建知识图谱的基础。针对电力计量领域需要，结合中文分词技术特点，基于联合学习思想，提出了一种基于联合学习的中文电力计量命名实体识别技术。该技术联合CNN-BLSTM-CRF模型与整合词典知识的分词模型，使其共享实体类别和置信度；同时将2个模型的先后计算顺序改为并行计算，减少了识别误差累积。结果表明，在不需要人工构建特征的情况下，方法的正确率、召回率、F值等均显著优于以往方法。     

从双层规划的角度看道德风险理论的一阶条件方法

本文首先对双层规划的一个特殊例子即道德风险模型中使用的一阶条件方法（FOA）做简要的梳理，然后提出一种更为一般的使FOA有效的原则与方法。新方法主要依赖于代理人对委托人设置的目标的最优反应映射是否存在不动点，这个性质不要求原问题与用一阶条件放松以后的问题之间的约束集等价，从而也不要求代理人的期望效用对行动具有全局凹性。在新方法下，可以用较为简单的方法证明FOA在以下两种情形之一有效，即如果分布函数是概率分布的凸组合或者分布函数来自某些特殊的指数族分布。