Minimum Message Length in Hybrid ARMA and LSTM Model Forecasting

Modeling and analysis of time series are important in applications including economics, engineering, environmental science and social science. Selecting the best time series model with accurate parameters in forecasting is a challenging objective for scientists and academic researchers. Hybrid models combining neural networks and traditional Autoregressive Moving Average (ARMA) models are being used to improve the accuracy of modeling and forecasting time series. Most of the existing time series models are selected by information-theoretic approaches, such as AIC, BIC, and HQ. This paper revisits a model selection technique based on Minimum Message Length (MML) and investigates its use in hybrid time series analysis. MML is a Bayesian information-theoretic approach and has been used in selecting the best ARMA model. We utilize the long short-term memory (LSTM) approach to construct a hybrid ARMA-LSTM model and show that MML performs better than AIC, BIC, and HQ in selecting the model—both in the traditional ARMA models (without LSTM) and with hybrid ARMA-LSTM models. These results held on simulated data and both real-world datasets that we considered.We also develop a simple MML ARIMA model.     

Whether the Support Region of Three-Bit Uniform Quantizer Has a Strong Impact on Post-Training Quantization for MNIST Dataset?

Driven by the need for the compression of weights in neural networks (NNs), which is especially beneficial for edge devices with a constrained resource, and by the need to utilize the simplest possible quantization model, in this paper, we study the performance of three-bit post-training uniform quantization. The goal is to put various choices of the key parameter of the quantizer in question (support region threshold) in one place and provide a detailed overview of this choice’s impact on the performance of post-training quantization for the MNIST dataset. Specifically, we analyze whether it is possible to preserve the accuracy of the two NN models (MLP and CNN) to a great extent with the very simple three-bit uniform quantizer, regardless of the choice of the key parameter. Moreover, our goal is to answer the question of whether it is of the utmost importance in post-training three-bit uniform quantization, as it is in quantization, to determine the optimal support region threshold value of the quantizer to achieve some predefined accuracy of the quantized neural network (QNN). The results show that the choice of the support region threshold value of the three-bit uniform quantizer does not have such a strong impact on the accuracy of the QNNs, which is not the case with two-bit uniform post-training quantization, when applied in MLP for the same classification task. Accordingly, one can anticipate that due to this special property, the post-training quantization model in question can be greatly exploited.     

All‐optical virtual private network in passive optical networks

We review recent progress on all‐optical virtual‐private‐network (VPN) schemes in passive optical networks (PONs). PON is a promising candidate in future access areas to provide broadband services with low cost. With all‐optical virtual private network (VPN) function, PON can support efficient internetworking among end users with dedicated optical channels, thus enabling guaranteed bandwidth and enhanced security at the physical layer. Here, we discuss and compare existing schemes of all‐optical VPNs in time‐division‐multiplexed (TDM) PONs, and also recently proposed schemes for deployment in wavelength‐division‐multiplexed (WDM) PONs and two‐stage TDM/WDM PONs.     

Analytical solutions for the slow neutron capture process of heavy element nucleosynthesis

In this paper, the network equation for the slow neutron capture process (s-process) of heavy element nucleosynthesis is investigated. Dividing the s-process network reaction chains into two standard forms and using the technique of matrix decomposition, a group of analytical solutions for the network equation are obtained. With the analytical solutions, a calculation for heavy element abundance of the solar system is carried out and the results are in good agreement with the astrophysical measurements.     

差分吸收光谱系统监测污染物浓度实时预测模型

差分吸收光谱(DOAS)法是一种有效的监测大气污染气体浓度的光学遥感方法,不仅有好的时间分辨率,而且测量灵敏度也很高。但是由于遥测系统处于复杂的大气环境中,各种干扰因素以及恶劣的气候条件,都会对系统产生影响。针对现有实时、在线监测差分吸收光谱系统中存在的不足,作者提出了一种基于改进Elman网络的实时预测模型。利用逐步回归筛选预测因子,不仅降低了预测网络的复杂程度,而且增强了系统的预测实时性。利用带自适应学习率的动态BP算法对改进的Elman网络进行训练,使预测系统能更好地辨识要预测的差分吸收光谱系统,该模型能较准确地对DOAS系统监测污染物数据进行实时跟踪监控,一定程度上弥补了遥测系统的不足。     

一种基于诱骗态的广域量子安全直接通信网络方案

提出了一种基于诱骗态的广域量子安全直接通信网络方案.在每一个局域网中设置一个服务器负责量子态的产生和测量,从而提高了通信距离;将诱骗态的思想引入量子安全直接通信,采用不同的强度发送光脉冲,能够克服光子数目分割攻击,从而提高通信的安全性;根据信道参量估计了不同通信距离的通过率,为信道编码提供了依据.对所提方案进行了安全性分析,结果表明此方案能够实现远距离量子安全直接通信.     

Evaluation of Transparent Optical Network Performance in Presence of Incoherent Homodyne Crosstalk

A simple method to evaluate the performance of transparent optical networks (TON) that takes into account the incoherent homodyne crosstalk, signal distortion, inter-symbol interference (ISI), amplified spontaneous emission (ASE) noise, and arbitrary optical and electrical filters is proposed. This method combines the effect of the incoherent homodyne crosstalk with an exhaustive Gaussian approximation for the influence of ASE noise, signal distortion, and ISI. It is shown that the incoherent homodyne crosstalk penalty depends on the ASE noise accumulated in the TON, ISI, and extinction ratio. The proposed method estimates are compared with experimental results published by other authors. Good agreement is observed for one, two, and four incoherent homodyne crosstalk signals, revealing that the proposed method provides accurate estimates even when the methods based on a Gaussian approximation for the incoherent homodyne crosstalk fail. The proposed method is used to assess the performance of a TON composed by optical add-drop multiplexers based on arrayed waveguide gratings. Results reveal that incoherent homodyne crosstalk, ASE noise, and ISI impair the TON performance and, therefore, none can be neglected in the TON design.     

利用神经网络提高偏最小二乘法的NIR多组分分析精度

提出了一种神经网络(ANN)和偏最小二乘法(PLS)结合的新的近红外(NIR)多组分分析法。该方法首先把训练样本中待测组分涵盖的浓度区间分成若干个子区间,利用各个子区间的训练样本分别建立PLS校正模型,然后利用ANN对未知样本进行分类,判断其所属的浓度子区间,应用对应子区间上的校正模型计算预测样本的组分浓度。和传统的PLS比较,此方法改善了模型的适应性,显著地提高了预测精度。实验及数据处理结果证明了本方法的有效性。     

基于神经网络的NMR谱图自动相位校正综合算法

目前,已有多种算法被应用在核磁共振谱图自动相位校正中,由于各种算法本身特性和所基于谱图的具体特性的差异,不同算法对于特征不同的谱图的适用性也各不相同. 针对这一情况,文在综合研究多种现有自动相位校正算法的基础上,提出了一种基于神经网络的,可以根据谱图的特征来选取最合适的算法进行自动相位校正的综合算法. 实验表明,本算法可以获得比以往方法更好的计算结果.     

Genetic algorithm-support vector regression for high reliability SHM system based on FBG sensor network

Structural Health Monitoring (SHM) based on fiber Bragg grating (FBG) sensor network has attracted considerable attention in recent years. However, FBG sensor network is embedded or glued in the structure simply with series or parallel. In this case, if optic fiber sensors or fiber nodes fail, the fiber sensors cannot be sensed behind the failure point. Therefore, for improving the survivability of the FBG-based sensor system in the SHM, it is necessary to build high reliability FBG sensor network for the SHM engineering application. In this study, a model reconstruction soft computing recognition algorithm based on genetic algorithm-support vector regression (GA-SVR) is proposed to achieve the reliability of the FBG-based sensor system. Furthermore, an 8-point FBG sensor system is experimented in an aircraft wing box. The external loading damage position prediction is an important subject for SHM system; as an example, different failure modes are selected to demonstrate the SHM system's survivability of the FBG-based sensor network. Simultaneously, the results are compared with the non-reconstruct model based on GA-SVR in each failure mode. Results show that the proposed model reconstruction algorithm based on GA-SVR can still keep the predicting precision when partial sensors failure in the SHM system; thus a highly reliable sensor network for the SHM system is facilitated without introducing extra component and noise.