Functional cardiotoxicity profiling and screening using the xCELLigence RTCA Cardio System

Cardiac safety testing of lead drug candidates is an important part of the drug discovery and development process. All new chemical entities need to be subjected to extensive preclinical assessment for cardiac liability, especially for a potentially fatal form of ventricular arrhythmia referred to as Torsades de Pointes. We have developed an innovative label-free, real-time system, the xCELLigence RTCA Cardio System, which is designed to monitor contractility of cardiomyocytes based on impedance measurement. The assay is performed using specially designed microtiter plates that are integrated with gold microelectrodes. The system was validated using mouse embryonic stem cell-derived cardiomyocytes, human-induced pluripotent stem cell-derived cardiomyocytes, and rat neonatal primary cardiomyocytes by applying a variety of tool compounds and drugs with known mechanisms of action. Our data show that the time resolution in the assay can provide important information about compound action. Furthermore, the impedance-based beating profile in response to compound treatment can provide mechanistic toxicity information regarding the target being modulated and may be able to flag pro-arrhythmic compounds. We believe the real-time and kinetic aspect of this technology combined with beat-to-beat measurement of cardiomyocyte contraction would make this instrument an important part of preclinical cardiac safety assessment.     

基于改进残差密集网络的心律失常自动分类

实现对不同类型心律失常的自动分类可为医生提供可靠诊断信息,有效提高该类疾病的诊断效率。因此,本文提出一种改进的残差密集网络用于心律失常的自动分类。该模型将改进的残差密集块作为残差密集网络的基本模块,使用深度可分离卷积替代传统卷积可有效提取通道间特征,降低参数量,同时引入通道注意力机制,实现特征选择,提高重要特征的权重分布。实验基于2018中国生理信号挑战赛提供的公开数据集,对9种类型的心律失常进行分类,F1均值达到81.2%,优于主流的深度学习网络模型。实验结果验证了该模型的可行性与优势,为心律失常分类提供了一种新的方法。     

低密度校验码在瑞利衰落信道中的性能分析

低密度校验（LDPC）码具有编码增益高，译码速度快，可并行译码等特点，是当前编码界的一个研究热点，但是目前已有的研究成果都集中在高斯信道上，该文分析和讨论了LDPC码在瑞利衰落信道下的性能，应用联合界技术推导了一个规则友的性能限，并给同了仿真结果，且发现LDPC码在瑞利衰落信道下也具有非常好的性能。     

A source of bias in the analysis of single channel data: assessing the apparent interaction between channel proteins.

A recent study of single sodium channel currents in neuroblastoma cells suggested interaction between ion channels in close proximity to one another (T. Kiss and K. Nagy, Eur. Biophys. J. 12, 13, 1985). The opening of one channel appeared to affect the likelihood that neighboring channels might open. Some of the conclusions were based on the analysis of observed channel openings that were segregated depending on whether one channel or more than one channel was open at the same time. We hypothesized that the longer one channel remained open, the more likely another channel operating independently, would open, thereby creating the impression of an apparent coupling of channel behavior. We performed simulations and measurements of single sodium channel currents to determine whether the technique of event segregation could account for apparent channel interactions. The simulations showed that the segregation of overlapping (more than one channel open at the same time) and nonoverlapping events led to a bias in the estimated open time and the derived closing rate. To avoid the bias, we found that random pairing of opening and closing events provided an unbiased estimate of the mean closing rate. Using this random assignment approach, we showed that the mean closing rate of single sodium channels in neonatal rat myocytes decreased with depolarization over a limited range of membrane potential. This suggested that the underlying closure mechanism(s) was voltage dependent. From the analysis of open times, we found no evidence for channel interaction in the time scale of tens of milliseconds. Depolarizing steps without events occurred in runs suggesting the existence of long-lived shut state(s). Double pulse experiments with the prepulse and test pulse above threshold showed significant inactivation of channels that did not open. The rate of inactivation of shut channels was substantially slower than the closure rate of open channels. The rate of inactivation of cardiac sodium channels appeared to be strongly dependent on the initial channel state.     

Heat transfer blockage in small scale combustion of polymers

Flame heat transfer blockage occurs as fuel vapors,soot and products of combustion near a burning fuel surface block much of the heat feedbacks(including external radiative heat flux)to the fuel surface of a burning object.Blockage clearly affects burning rates and heat release rates of fires.This needs to be included when calculating flame heat transfer in fire growth models.An understanding of burning of materials in small scale fires is of broad and vital importance for predicting their burning performan...     

通道与空间注意力图像超分辨率网络

单幅图像超分辨率（Single Image Super Resolution,SISR）在计算机视觉领域占有重要地位,该技术旨在从低分辨率图像中重建出高分辨率图像。近年来,深度神经网络在SISR领域起到了至关重要的作用,然而,目前利用卷积神经网络平等地对待高频与低频特征,使得高频细节的重建表现不佳,输出过于平滑,缺少纹理信息。另一方面,过于深的网络不容易收敛,并且随着神经网络的深度增长,来自前一层的长期信息很容易在后期层中减弱或丢失,使得重建收益不能正比于网络的深度与计算复杂度。针对以上问题,对用于SISR的卷积神经网络的基本块提出了空间注意力模块与通道注意力模块,在同一通道中,不同位置的信息被空间注意力模块赋予不同的权重,不同通道间的权重由通道注意力模块决定,这使得高频信息在重建任务中获得更高的地位,提高了重建指标。进一步地提出了长期特征调制模块将网络的层深度转化为块深度,大大缩小了网络深度,以解决前层长期信息的丢失问题。在Set5等多个基准数据集上的峰值信噪比（PSNR）均比目前其他基于深度卷积神经网络的方法有所提升,这证明了提出的方法的有效性与先进性。     

Turbo equalization for block fading MIMO channels using random signal mapping

We present a novel turbo equalization scheme for block fading MIMO channels, where diversity is achieved using random signal mapping. We show that the computational complexity of the proposed scheme is not a function of the number of transmit antennas, and compares favorably to the complexity of similar systems based on space-time trellis codes and space-time block codes. Finally, we provide simulation results showing that the proposed scheme achieves full diversity, and investigate its performance in terms of the number of transmit antennas, ISI length and imperfect channel knowledge.     

正交频分复用通信系统中的快变信道估计

信道的快速变化导致正交频分复用通信系统(OFDM)载波间的正交性遭到破坏,从而产生载波间的相互干扰(ICI),传统的信道估计方法(不可虑ICI影响)不再适用。为此,Choi等人提出了相应的快变信道估计方法,但该方法计算量较大,且需要信道的统计特性知识,而实际信道的统计特性往往无法精确得到。Linartz等人证明:在大多数应用条件下,信道在一个OFDM帧周期内的变化近似满足线性特性。对OFDM系统的ICI特性进行了时域和频域两方面分析,并利用信道变化的线性特性,提出了一种新的快变信道估计方法,该方法具有计算量小、且不需要信道的统计特性信息的优点。文章还得到如下结论:与慢变信道情形不同,在信道快速变化条件下,当OFDM系统中的训练序列应分成许多长度相同的组,且这些组在OFDM帧中等间隔分布时,信道估计误差最小。     

Challenges in Computational Modeling of Two-Phase Transport in Polymer Electrolyte Fuel Cells Flow Channels: A Review

Excess of liquid water in gas channels of polymer electrolyte fuel cells is responsible for malfunctioning of these devices. Not only it decreases their efficiency via partial blockage of reactants and pressure drop, but it can also lead to the irreversible damage due to oxygen starvation in case of complete channel flooding or full coverage of the gas diffusion layer with a liquid film. Liquid water evacuation is carried out via airflow in gas channels. Several experimental and computational techniques have been applied to date for the analysis of the coupled airflow–water behavior in order to understand the impact of fuel cell design and operation regimes upon the liquid water accumulation. Considerable progress has been achieved with the development of sophisticated computational fluid dynamics (CFD) tools. Nevertheless, the complexity of the problem under consideration leaves several issues unresolved. In this paper, analysis techniques applied to liquid water–airflow transport in fuel cells gas channels are reviewed and most important results are summarized. Computationally efficient, yet strongly simplified analytical models are discussed. Afterwards, CFD approaches including the conventional fixed grid (Eulerian) and the novel embedded Eulerian–Lagrangian models are described. Critical comparative assessment of the existing methods is provided at the end of the paper and the unresolved issues are highlighted.     

车联网中多跳广播转发节点选择研究

在车载自组织网络中(VANET),大部分应用采用多跳广播来交换道路交通信息,其中最典型的为泛洪广播,但是随着车辆和节点的增多泛洪广播会引起广播风暴,导致信道阻塞。结合模糊逻辑,提出一种基于接收节点的多跳广播方案,依靠接收节点的覆盖范围、信道忙闲比和移动因素,由模糊逻辑系统决定当前节点是否具备转发资格。通过于其它方案的仿真对比验证,所提出的方案在投递率与时延方面具有更好的性能,可有效提高信道利用率,降低广播风暴风险。     

Modeling the binding modes of Kv1.5 potassium channel and blockers

The ultra-rapid delayed rectifier potassium current (I(Kur)), encoded by Kv1.5 gene, is the critical determinant of Phase I repolarization of action potential duration (APD). The evidences that Kv1.5 gene expresses more extensively in human atrial myocytes than in ventricle and the I(Kur) currents has not been recorded in the human ventricle, suggest Kv1.5 potassium channel as a selective target for the treatment of atrial fibrillation (AF). Recent mutagenesis studies have provided us some evidences that are useful in designing Kv1.5 blockers. In order to further evaluate these molecular biological information, the homology model of Kv1.5 potassium channel was established based on the Kv1.2 crystal structure (PDB entry: 2A79) using MODELLER 9v2 program. After the molecular dynamics refinement, the optimized homology model was assessed as a reliable structure by PROCHECK, ERRAT, WHAT-IF, PROSA2003 and DOPE graph. The results of molecular docking studies on different Kv1.5 inhibitors are in agreement with the published mutagenesis data. Based on the docking conformations, a pharmacophore model was developed by HipHop algorithm in order to probe the common features of blockers. By analyzing the results, active site architecture, certain key residues and pharmacophore common-features that are responsible for substrate specificity were identified on the Kv1.5 potassium channel, which would be very helpful in understanding the blockade mechanism of Kv1.5 potassium channel and providing insights into rational design of novel Kv1.5 blockers.     

Identification of sparse impulse responses – design and implementation using the partial Haar block wavelet transform

This paper proposes an implementation for identifying sparse impulse responses. The new scheme follows the approach in which the location of the channel response peak is estimated in the wavelet domain. A short time-domain adaptive filter is then located about the estimated peak to identify the sparse response. The primary purpose of this paper is to present an efficient design of such a system. The use of a new block wavelet transform results in up to 70% less computational complexity and improved peak detection, as compared to previous solutions. A new robust time-domain adaptive filtering location and update scheme is also proposed that significantly reduces the occurrence of jitter problems and leads to improved residual mean-square error performance. The behavior of the transform-domain adaptive filter is analyzed, the Wiener solution is determined, and an accurate analytical model is obtained for the mean-square deviation of the adaptive coefficients. Monte Carlo simulations show excellent echo cancellation performance for typical ITU-T echo channels.     

基于注意力机制和多空间金字塔池化的实时目标检测算法

YOLOv4计算复杂度高、空间金字塔池化模块仅一次增强特征融合网络的深层区域特征图的表征能力、检测头网络的特征图难以突出重要通道特征;针对以上问题,提出一种基于注意力机制和多空间金字塔池化的实时目标检测算法;该算法采用多空间金字塔池化,提取局部特征和全局特征,融合多重感受野,加强特征融合网络的浅、中、深层特征图的表征能力;引入压缩激励通道注意力机制,建模通道间的相关性,自适应调整特征图各个通道的权重,从而使网络更加关注重要特征;特征融合和检测头网络中使用深度可分离卷积,减少了网络参数量;实验结果表明,所提算法的均值平均精度均高于其他七种主流对比算法;与YOLOv4相比,参数量、模型大小分别减少了27.85 M和106.25 MB,所提算法在降低复杂度的同时,提高了检测准确度;且该算法的检测速率达到33.70 帧/秒,满足实时性要求。     

快速DCT域分形图像编码方法

分形图像压缩编码复杂度较高,而ＤＣＴ域的分形图像压缩方法虽然在性能上有所提高,但是对每个图像块的处理中又分别增加了一次二维ＤＣＴ,使得编码复杂度进一步提高,为降低ＤＣＴ域分形图像压缩的编码复杂度,引入平坦块的概念,对那些经过ＤＣＴ后,能量非常集中的块不做块匹配,而是直接编码输出。实验结果表明,与一般的ＤＣＴ域分形图像压缩方法相比,这种方法在提高编码速度的同时,也降低了比特率。     

A Drug Candidate Design Environment Using Evolutionary Computation

This paper describes the candidate design environment we developed for efficient identification of promising drug candidates. Developing effective drugs from active molecules is a challenging problem which requires the simultaneous satisfaction of many factors. Traditionally, the drug discovery process is conducted by medicinal chemists whose vital expertise is not readily quantifiable. Recently, in silico modeling and virtual screening have been emerging as valuable tools despite their mixed results early on. Our approach combines the capabilities of computational models with human knowledge using a genetic algorithm and interactive evolutionary computation. We enable the chemist's expertise to play a key role in every stage of the discovery process. Our evolved structures are guaranteed to be within the chemistry space specified by the medicinal chemist, thereby making the results plausible. In this paper, we describe our approach, introduce a case study to test our methodology, and present our results.     

Effective Generation of Data Broadcast Schedules with Different Allocation Numbers for Multiple Wireless Channels

Existing methods of scheduling data items over multiple wireless broadcast channels focus on the assignment of a data item to a channel. However, data items are not allocated more than once per broadcast cycle to a single channel. Our scheme considers the numbers of copies of a data item that should be allocated in the context of the channel assignment problem and aims to reduce the average data access time by allocating a popular data item more than once per cycle to the channel to which it is assigned. The number of times that each data item is allocated reflects its access probability. Simulation results show that our scheme reduces the average expected delay, especially when there are few channels.     

BLOCK DECOUPLING CONTROL OF LINEAR MULTIVARIABLE SYSTEMS

This paper gives parameterizations of block decoupling controllers and achievable block-diagonal I/O maps for linear multivariable plants under unity-feedback configuration. The parameterizations, based on coprime factorizations and a special generalized Bezout identity, do not require the computation of Smith McMillan forms and, thus, can be reliably computed. It is also shown that through the construction of an admissible block decoupling precompensator, block decoupling controller design reduces to a set of controller designs for the block channels which are smaller in dimension.     

基于通道性能度量的神经网络结构搜索算法

卷积神经网络的表征与预测能力往往依赖结构合理性,但其主流结构均由人工设计,存在设计难度高、算力要求强、时间开销大等问题.如何让神经网络自主搜索合理结构并节约计算资源是当前的研究重点.目前,基于部分通道连接的可微分结构搜索算法,以其高效的显存利用率在搜索速度和分类性能上表现良好.然而,其针对通道的随机采样策略易造成重要信息丢失,当通道连接不足时性能明显下降.为此,提出一种基于通道性能度量的神经网络结构搜索算法,利用注意力机制提取通道重要性系数,并以此对通道进行排序采样.此外,考虑到预热阶段导致搜索不充分,产生较大离散化误差,在结构权重连续化的过程中设计温度正则化系数,提升权重差异.实验表明,所提算法能够在节约计算资源的基础上搜索出更优的卷积神经网络结构.