Adaptive Partial Task Offloading and Virtual Resource Placement in SDN/NFV-Based Network Softwarization

Edge intelligence brings the deployment of applied deep learning(DL)models in edge computing systems to alleviate the core backbone network congestions.The setup of programmable software-defined networking(SDN)control and elastic virtual computing resources within network functions virtualization(NFV)are cooperative for enhancing the applicability of intelligent edge softwarization.To offer advancement for multi-dimensional model task offloading in edge networks with SDN/NFV-based control softwarization,this study proposes a DL mechanism to recommend the optimal edge node selection with primary features of congestion windows,link delays,and allocatable bandwidth capacities.Adaptive partial task offloading policy considered the DL-based recommendation to modify efficient virtual resource placement for minimizing the completion time and termination drop ratio.The optimization problem of resource placement is tackled by a deep reinforcement learning(DRL)-based policy following the Markov decision process(MDP).The agent observes the state spaces and applies value-maximized action of available computation resources and adjustable resource allocation steps.The reward formulation primarily considers taskrequired computing resources and action-applied allocation properties.With defined policies of resource determination,the orchestration procedure is configured within each virtual network function(VNF)descriptor using topology and orchestration specification for cloud applications(TOSCA)by specifying the allocated properties.The simulation for the control rule installation is conducted using Mininet and Ryu SDN controller.Average delay and task delivery/drop ratios are used as the key performance metrics.

LaNets:Hybrid Lagrange Neural Networks for Solving Partial Differential Equations

We propose new hybrid Lagrange neural networks called LaNets to predict the numerical solutions of partial differential equations.That is,we embed Lagrange interpolation and small sample learning into deep neural network frameworks.Concretely,we first perform Lagrange interpolation in front of the deep feedforward neural network.The Lagrange basis function has a neat structure and a strong expression ability,which is suitable to be a preprocessing tool for pre-fitting and feature extraction.Second,we introduce small sample learning into training,which is beneficial to guide themodel to be corrected quickly.Taking advantages of the theoretical support of traditional numerical method and the efficient allocation of modern machine learning,LaNets achieve higher predictive accuracy compared to the state-of-the-artwork.The stability and accuracy of the proposed algorithmare demonstrated through a series of classical numerical examples,including one-dimensional Burgers equation,onedimensional carburizing diffusion equations,two-dimensional Helmholtz equation and two-dimensional Burgers equation.Experimental results validate the robustness,effectiveness and flexibility of the proposed algorithm.

TCAS-PINN:Physics-informed neural networks with a novel temporal causality-based adaptive sampling method

Physics-informed neural networks(PINNs)have become an attractive machine learning framework for obtaining solutions to partial differential equations(PDEs).PINNs embed initial,boundary,and PDE constraints into the loss function.The performance of PINNs is generally affected by both training and sampling.Specifically,training methods focus on how to overcome the training difficulties caused by the special PDE residual loss of PINNs,and sampling methods are concerned with the location and distribution of the sampling points upon which evaluations of PDE residual loss are accomplished.However,a common problem among these original PINNs is that they omit special temporal information utilization during the training or sampling stages when dealing with an important PDE category,namely,time-dependent PDEs,where temporal information plays a key role in the algorithms used.There is one method,called Causal PINN,that considers temporal causality at the training level but not special temporal utilization at the sampling level.Incorporating temporal knowledge into sampling remains to be studied.To fill this gap,we propose a novel temporal causality-based adaptive sampling method that dynamically determines the sampling ratio according to both PDE residual and temporal causality.By designing a sampling ratio determined by both residual loss and temporal causality to control the number and location of sampled points in each temporal sub-domain,we provide a practical solution by incorporating temporal information into sampling.Numerical experiments of several nonlinear time-dependent PDEs,including the Cahn–Hilliard,Korteweg–de Vries,Allen–Cahn and wave equations,show that our proposed sampling method can improve the performance.We demonstrate that using such a relatively simple sampling method can improve prediction performance by up to two orders of magnitude compared with the results from other methods,especially when points are limited.

Data-Driven Modeling of Partially Observed Biological Systems

We present a numerical approach for modeling unknown dynamical systems using partially observed data,with a focus on biological systems with(relatively)complex dynamical behavior.As an extension of the recently developed deep neural network(DNN)learning methods,our approach is particularly suitable for practical situations when(i)measurement data are available for only a subset of the state variables,and(ii)the system parameters cannot be observed or measured at all.We demonstrate that,with a properly designed DNN structure with memory terms,effective DNN models can be learned from such partially observed data containing hidden parameters.The learned DNN model serves as an accurate predictive tool for system analysis.Through a few representative biological problems,we demonstrate that such DNN models can capture qualitative dynamical behavior changes in the system,such as bifurcations,even when the parameters controlling such behavior changes are completely unknown throughout not only the model learning process but also the system prediction process.The learned DNN model effectively creates a“closed”model involving only the observables when such a closed-form model does not exist mathematically.

Distributed wireless quantum communication networks with partially entangled pairs

Wireless quantum communication networks transfer quantum state by teleportation. Existing research focuses on maximal entangled pairs. In this paper, we analyse the distributed wireless quantum communication networks with partially entangled pairs. A quantum routing scheme with multi-hop teleportation is proposed. With the proposed scheme, is not necessary for the quantum path to be consistent with the classical path. The quantum path and its associated classical path are established in a distributed way. Direct multi-hop teleportation is conducted on the selected path to transfer a quantum state from the source to the destination. Based on the feature of multi-hop teleportation using partially entangled pairs, if the node number of the quantum path is even, the destination node will add another teleportation at itself. We simulated the performance of distributed wireless quantum communication networks with a partially entangled state. The probability of transferring the quantum state successfully is statistically analyzed. Our work shows that multi-hop teleportation on distributed wireless quantum networks with partially entangled pairs is feasible,

Mobility-Aware Partial Computation Offloading in Vehicular Networks: A Deep Reinforcement Learning Based Scheme

Encouraged by next-generation networks and autonomous vehicle systems,vehicular networks must employ advanced technologies to guarantee personal safety,reduce traffic accidents and ease traffic jams.By leveraging the computing ability at the network edge,multi-access edge computing(MEC)is a promising technique to tackle such challenges.Compared to traditional full offloading,partial offloading offers more flexibility in the perspective of application as well as deployment of such systems.Hence,in this paper,we investigate the application of partial computing offloading in-vehicle networks.In particular,by analyzing the structure of many emerging applications,e.g.,AR and online games,we convert the application structure into a sequential multi-component model.Focusing on shortening the application execution delay,we extend the optimization problem from the single-vehicle computing offloading(SVCOP)scenario to the multi-vehicle computing offloading(MVCOP)by taking multiple constraints into account.A deep reinforcement learning(DRL)based algorithm is proposed as a solution to this problem.Various performance evaluation results have shown that the proposed algorithm achieves superior performance as compared to existing offloading mechanisms in deducing application execution delay.

DISTRIBUTED PARAMETER NEURAL NETWORKS FOR SOLVING PARTIAL DIFFERENTIAL EQUATIONS

Novel distributed parameter neural networks are proposed for solving partial differential equations, and their dynamic performances are studied in Hilbert space. The locally connected neural networks are obtained by separating distributed parameter neural networks. Two simulations are also given. Both theoretical and computed results illustrate that the distributed parameter neural networks are effective and efficient for solving partial differential equation problems.

Prediction of Partial Ring Current Index Using LSTM Neural Network

The local time dependence of the geomagnetic disturbances during magnetic storms indicates the necessity of forecasting the localized magnetic storm indices.For the first time,we construct prediction models for the SuperMAG partial ring current indices(SMR-LT),with the advance time increasing from 1 h to 12 h by Long Short-Term Memory(LSTM)neural network.Generally,the prediction performance decreases with the advance time and is better for the SMR-06 index than for the SMR-00,SMR-12,and SMR-18 index.For the predictions with 12 h ahead,the correlation coefficient is 0.738,0.608,0.665,and 0.613,respectively.To avoid the over-represented effect of massive data during geomagnetic quiet periods,only the data during magnetic storms are used to train and test our models,and the improvement in prediction metrics increases with the advance time.For example,for predicting the storm-time SMR-06 index with 12 h ahead,the correlation coefficient and the prediction efficiency increases from 0.674 to 0.691,and from 0.349 to 0.455,respectively.The evaluation of the model performance for forecasting the storm intensity shows that the relative error for intense storms is usually less than the relative error for moderate storms.

Development a Spectrophotometric of Fe(Ⅲ), Al(Ⅲ) and Cu(Ⅱ) Using Eriochrome Cyanine R Ligand and Assessment of the Obtained Data by Partial Least-Squares and Artificial Neural Network Method-Application to Natural Waters

Simultaneous determination of heavy metal cations and accurate quantitative prediction of them are of great interest in analytical chemistry.This work has focused on a comprehensive comparison of partial least squares(PLS-1)and artificial neural networks(ANN)as two types of chemometric methods.For this purpose,aluminum,iron and copper were studied as three analytes whose UV-Vis absorption spectra highly overlap each other.Accordance with determined parameters(ligand concentration,pH,waiting times,the relationship between absorbance and concentration of metal ion effect and foreign ions)are provided and the optimum conditions.After establishing the optimum conditions for Fe^(3+),Al^(3+) and Cu^(2+) containing mixtures spectrophotometric determinations and the data calibration method of least squares(PLS-1)regression,and artificial neural network(ANN)methods were used.Chemometric methods are applied in a fast,simple,and the results are applicable.

Quantum communication for satellite-to-ground networks with partially entangled states

To realize practical wide-area quantum communication,a satellite-to-ground network with partially entangled states is developed in this paper.For efficiency and security reasons,the existing method of quantum communication in distributed wireless quantum networks with partially entangled states cannot be applied directly to the proposed quantum network.Based on this point,an efficient and secure quantum communication scheme with partially entangled states is presented.In our scheme,the source node performs teleportation only after an end-to-end entangled state has been established by entanglement swapping with partially entangled states.Thus,the security of quantum communication is guaranteed.The destination node recovers the transmitted quantum bit with the help of an auxiliary quantum bit and specially defined unitary matrices.Detailed calculations and simulation analyses show that the probability of successfully transferring a quantum bit in the presented scheme is high.In addition,the auxiliary quantum bit provides a heralded mechanism for successful communication.Based on the critical components that are presented in this article an efficient,secure,and practical wide-area quantum communication can be achieved.

Application of neural network model coupling with the partial least-squares method for forecasting watre yield of mine

Scientific forecasting water yield of mine is of great significance to the safety production of mine and the colligated using of water resources. The paper established the forecasting model for water yield of mine, combining neural network with the partial least square method. Dealt with independent variables by the partial least square method, it can not only solve the relationship between independent variables but also reduce the input dimensions in neural network model, and then use the neural network which can solve the non-linear problem better. The result of an example shows that the prediction has higher precision in forecasting and fitting.

A Partially Non-Cryptographic Security Routing Protocol in Mobile Ad Hoc Networks

In this paper, we propose a partially non-cryptographic security routing protocol （PNCSR） that protects both routing and data forwarding operations through the same reactive approach. PNCSR only apply public-key cryptographic system in managing token, but it doesn＇t utilize any cryptographic primitives on the routing messages. In PNCSR, each node is fair. Local neighboring nodes collaboratively monitor each other and sustain each other. It also uses a novel credit strategy which additively increases the token lifetime each time a node renews its token. We also analyze the storage, computation, and communication overhead of PNCSR, and provide a simple yet meaningful overhead comparison. Finally, the simulation results show the effectiveness of PNCSR in various situations.

A Highly Compatible Circular-Shifting Network for Partially Parallel QC-LDPC Decoder

The conventional methodology for designing QC-LDPC decoders is applied for fixed configurations used in wireless communication standards, and the supported largest expansion factor Z (the parallelism of the layered decoding) is a fixed number. In this paper, we study the circular-shifting network for decoding LDPC codes with arbitrary Z factor, especially for decoding large Z (Z P) codes, where P is the decoder parallelism. By buffering the P-length slices from the memory, and assembling the shifted slices in a fixed routine, the P-parallelism shift network can process Z-parallelism circular-shifting tasks. The implementation results show that the proposed network for arbitrary sized data shifting consumes only one times of additional resource cost compared to the traditional solution for only maximum P sized data shifting, and achieves significant saving on area and routing complexity.

ESR-PINNs:Physics-informed neural networks with expansion-shrinkage resampling selection strategies

Neural network methods have been widely used in many fields of scientific research with the rapid increase of computing power.The physics-informed neural networks(PINNs)have received much attention as a major breakthrough in solving partial differential equations using neural networks.In this paper,a resampling technique based on the expansion-shrinkage point(ESP)selection strategy is developed to dynamically modify the distribution of training points in accordance with the performance of the neural networks.In this new approach both training sites with slight changes in residual values and training points with large residuals are taken into account.In order to make the distribution of training points more uniform,the concept of continuity is further introduced and incorporated.This method successfully addresses the issue that the neural network becomes ill or even crashes due to the extensive alteration of training point distribution.The effectiveness of the improved physics-informed neural networks with expansion-shrinkage resampling is demonstrated through a series of numerical experiments.

A dive into spectral inference networks: improved algorithms for self-supervised learning of continuous spectral representations

We propose a self-supervising learning framework for finding the dominant eigenfunction-eigenvalue pairs of linear and self-adjoint operators.We represent target eigenfunctions with coordinate-based neural networks and employ the Fourier positional encodings to enable the approximation of high-frequency modes.We formulate a self-supervised training objective for spectral learning and propose a novel regularization mechanism to ensure that the network finds the exact eigenfunctions instead of a space spanned by the eigenfunctions.Furthermore,we investigate the effect of weight normalization as a mechanism to alleviate the risk of recovering linear dependent modes,allowing us to accurately recover a large number of eigenpairs.The effectiveness of our methods is demonstrated across a collection of representative benchmarks including both local and non-local diffusion operators,as well as high-dimensional time-series data from a video sequence.Our results indicate that the present algorithm can outperform competing approaches in terms of both approximation accuracy and computational cost.

PSO-DBNet for Peak-to-Average Power Ratio Reduction Using Deep Belief Network

Data transmission through a wireless network has faced various signal problems in the past decades.The orthogonal frequency division multiplexing(OFDM)technique is widely accepted in multiple data transfer patterns at various frequency bands.A recent wireless communication network uses OFDM in longterm evolution(LTE)and 5G,among others.The main problem faced by 5G wireless OFDM is distortion of transmission signals in the network.This transmission loss is called peak-to-average power ratio(PAPR).This wireless signal distortion can be reduced using various techniques.This study uses machine learning-based algorithm to solve the problem of PAPR in 5G wireless communication.Partial transmit sequence(PTS)helps in the fast transfer of data in wireless LTE.PTS is merged with deep belief neural network(DBNet)for the efficient processing of signals in wireless 5G networks.Result indicates that the proposed system outperforms other existing techniques.Therefore,PAPR reduction in OFDM by DBNet is optimized with the help of an evolutionary algorithm called particle swarm optimization.Hence,the specified design supports in improving the proposed PAPR reduction architecture.

突发事件下列车开行方案调整对铁路客运网络鲁棒性的影响

面对突发事件的发生,铁路会采取列车开行方案调整措施进行应对以减少突发事件带来的危害,极大地影响了铁路客运网络的鲁棒性。结合突发事件对铁路客运网络的影响区域性特点,选取部分停运的列车开行方案调整措施,确定突发事件对铁路客运网络的扰动,通过构建鲁棒性指标,进行了列车开行方案调整对铁路客运网络鲁棒性的仿真设计。以辽宁省铁路客运网络为例,计算不同比例停运下铁路客运网络的鲁棒性,研究发现:2种网络指标的度量下,最大连通子图的鲁棒性要好于可达性。列车部分停运下,当列车停运比例大于70%或车站节点数大于10时,会引起最大连通子图指标的快速下降;当列车停运比例大于10%或车站节点数大于9时,会引起可达性指标的快速下降。

基于自适应神经网络的PDEs求解研究

针对当前基于神经网络的PDEs求解方法效率和精度均不够理想的缺陷,研究提出一种基于改进BP神经网络(BP neural network,BPNN)的PDEs求解模型。首先,参照自适应网格法来改进神经网络结构,构建自适应神经网络,改进模型的输出精度;其次,提出一种引入Levy飞行机制和鲸鱼优化算法(Whale optimization algorithm,WOA)的改进海鸥优化算法(Improved Seagull Optimization Algorithm,ISOA)来优化BPNN,寻找BPNN的最佳参数,提高模型的性能;基于上述内容,构建基于ISOA-BPNNPDEs智能求解模型。结果显示,该模型的F1值为95.74%,准确率达到97.96%,输出误差为0.021,优于当前最先进的两种PDEs求解模型。上述内容表明,研究构建的基于ISOA-BPNNPDEs智能求解模型能够高效、准确地实现PDEs求解,为PDEs求解研究提供了新的路径。

面向轻量化的改进YOLOv7棉杂检测算法

针对棉纺厂原棉吞吐量大、检测时间长而常见卷积神经网络无法实现高实时检测的问题,提出基于轻量化改进的YOLOv7模型对原棉杂质的检测算法,旨在快速高效地对棉杂质进行检测。首先通过删减YOLOv7模型冗余的卷积层从而提高检测速度;其次在主干网络内添加FasterNet卷积降低模型的计算负担,减少特征图的冗余性,实现高实时检测;最后在颈部网络内运用CSP-RepFPN(Cross Stage Partial networks with Replicated Feature Pyramid Network)重构特征金字塔,增加特征信息流通,减少特征损失,提高检测精度。实验结果表明:在自建棉杂数据集上改进的YOLOv7模型在棉杂检测精度上达到了96.0%,检测时间比YOLOv7减少了37.5%;在公开DWC(Drinking Waste Classification)数据集上整体精度达到82.5%,检测时间仅为29.8 ms。改进的YOLOv7模型能够为原棉杂质的实时检测和识别分类提供一种轻量化的检测方法,大幅节约了时间成本。

A New Method for Solving Nonlinear Partial Differential Equations Based on Liquid Time-Constant Networks

In this paper,physics-informed liquid networks(PILNs)are proposed based on liquid time-constant networks(LTC)for solving nonlinear partial differential equations(PDEs).In this approach,the network state is controlled via ordinary differential equations(ODEs).The significant advantage is that neurons controlled by ODEs are more expressive compared to simple activation functions.In addition,the PILNs use difference schemes instead of automatic differentiation to construct the residuals of PDEs,which avoid information loss in the neighborhood of sampling points.As this method draws on both the traveling wave method and physics-informed neural networks(PINNs),it has a better physical interpretation.Finally,the KdV equation and the nonlinear Schr¨odinger equation are solved to test the generalization ability of the PILNs.To the best of the authors’knowledge,this is the first deep learning method that uses ODEs to simulate the numerical solutions of PDEs.