An Edge Server Placement Method Based on Reinforcement Learning

In mobile edge computing systems, the edge server placement problem is mainly tackled as a multi-objective optimization problem and solved with mixed integer programming, heuristic or meta-heuristic algorithms, etc. These methods, however, have profound defect implications such as poor scalability, local optimal solutions, and parameter tuning difficulties. To overcome these defects, we propose a novel edge server placement algorithm based on deep q-network and reinforcement learning, dubbed DQN-ESPA, which can achieve optimal placements without relying on previous placement experience. In DQN-ESPA, the edge server placement problem is modeled as a Markov decision process, which is formalized with the state space, action space and reward function, and it is subsequently solved using a reinforcement learning algorithm. Experimental results using real datasets from Shanghai Telecom show that DQN-ESPA outperforms state-of-the-art algorithms such as simulated annealing placement algorithm (SAPA), Top-K placement algorithm (TKPA), K-Means placement algorithm (KMPA), and random placement algorithm (RPA). In particular, with a comprehensive consideration of access delay and workload balance, DQN-ESPA achieves up to 13.40% and 15.54% better placement performance for 100 and 300 edge servers respectively.     

Metaheuristic Algorithms Based on Compromise Programming for the Multi-Objective Urban Shipment Problem

The Vehicle Routing Problem (VRP) and its variants are found in many fields, especially logistics. In this study, we introduced an adaptive method to a complex VRP. It combines multi-objective optimization and several forms of VRPs with practical requirements for an urban shipment system. The optimizer needs to consider terrain and traffic conditions. The proposed model also considers customers’ expectations and shipper considerations as goals, and a common goal such as transportation cost. We offered compromise programming to approach the multi-objective problem by decomposing the original multi-objective problem into a minimized distance-based problem. We designed a hybrid version of the genetic algorithm with the local search algorithm to solve the proposed problem. We evaluated the effectiveness of the proposed algorithm with the Tabu Search algorithm and the original genetic algorithm on the tested dataset. The results show that our method is an effective decision-making tool for the multi-objective VRP and an effective solver for the new variation of VRP.     

A Multi-Objective Multi-Label Feature Selection Algorithm Based on Shapley Value

Multi-label learning is dedicated to learning functions so that each sample is labeled with a true label set. With the increase of data knowledge, the feature dimensionality is increasing. However, high-dimensional information may contain noisy data, making the process of multi-label learning difficult. Feature selection is a technical approach that can effectively reduce the data dimension. In the study of feature selection, the multi-objective optimization algorithm has shown an excellent global optimization performance. The Pareto relationship can handle contradictory objectives in the multi-objective problem well. Therefore, a Shapley value-fused feature selection algorithm for multi-label learning (SHAPFS-ML) is proposed. The method takes multi-label criteria as the optimization objectives and the proposed crossover and mutation operators based on Shapley value are conducive to identifying relevant, redundant and irrelevant features. The comparison of experimental results on real-world datasets reveals that SHAPFS-ML is an effective feature selection method for multi-label classification, which can reduce the classification algorithm’s computational complexity and improve the classification accuracy.     

量子势阱粒子群优化算法的改进研究

为提高量子势阱粒子群优化算法的优化能力, 通过分析目前量子势阱粒子群优化算法的设计过程, 提出了改进的量子势阱粒子群优化算法. 首先, 分别基于Delta势阱、谐振子和方势阱 提出了改进的量子势阱粒子群优化算法, 并提出了基于统计量均值的控制参数设计方法. 然后, 在势阱中心的设计方面, 为强调全局最优粒子的指导作用, 提出了基于自身最优粒子加权平均和动态随机变量的两种设计策略. 实验结果表明, 三种势阱粒子群优化算法性能比较接近, 都优于原算法, 且Delta势阱模型略优于其他两种.     

Design optimization of broadband extreme ultraviolet polarizer in high-dimensional objective space

With the purpose of designing the extreme ultraviolet polarizer with many objectives, a combined application of multi-objective genetic algorithms is theoretically proposed. Owing to the multi-objective genetic algorithm, the relationships between different designing objectives of extreme ultraviolet polarizer have been obtained by analyzing the distribution of nondominated solutions in the four-dimensional objective space, and the optimized multilayer design can be obtained by guiding the searching in the desired region based on the multi-objective genetic algorithm with reference direction. Compared with the conventional method of multilayer design, our method has a higher probability of achieving the optimal multilayer design. Our work should be the first research in optimizing the optical multilayer designs in the high-dimensional objective space, and our results demonstrate a potential application of our method in the designs of optical thin films.     

目标函数设置对放疗逆向计划多目标优化过程的影响

目前放疗逆向计划中常用的目标函数有两种:基于剂量分布的目标函数和基于剂量．体积直方图（DVH）的目标函数。实际系统都基于单目标优化算法进行,不考虑逆向计划的多目标性。在Pareto多目标优化理论的基础上,研究逆向计划的目标函数设置问题,比较基于剂量分布的目标函数和基于DVH的目标函数对多目标优化过程的影响,包括优化时间、收敛性和存在的问题等,为逆向计划过程中多目标优化目标函数的设置提供依据。There are two kinds of objective functions in radiotherapy inverse planning: dose distribution-based and Dose-Volume Histogram （DVH）-based functions. The treatment planning in our days is still a trial and error process because the multi-objective problem is solved by transforming it into a single objective problem using a specific set of weights for each object. This work investigates the problem of objective function setting based on Pareto multi-optimization theory, and compares the effect on multi-objective inverse planning of those two kinds of objective functions including calculation time, converge speed, etc. The basis of objective function setting on inverse planning is discussed.     

A Soar-Based Space Exploration Algorithm for Mobile Robots

Space exploration is a hot topic in the application field of mobile robots. Proposed solutions have included the frontier exploration algorithm, heuristic algorithms, and deep reinforcement learning. However, these methods cannot solve space exploration in time in a dynamic environment. This paper models the space exploration problem of mobile robots based on the decision-making process of the cognitive architecture of Soar, and three space exploration heuristic algorithms (HAs) are further proposed based on the model to improve the exploration speed of the robot. Experiments are carried out based on the Easter environment, and the results show that HAs have improved the exploration speed of the Easter robot at least 2.04 times of the original algorithm in Easter, verifying the effectiveness of the proposed robot space exploration strategy and the corresponding HAs.     

Application of generalized complex modes to the calculation of the forced response of three-dimensional poroelastic materials

Finite element models based on Biot's {u,P} formulation for poroelastic materials are widely used to predict the behaviour of structures involving porous media. The numerical solution of such problems requires however important computational resources and the solution algorithms are not optimized. To improve the solution process, a modal approach based on an extension of the complex modes technique has been proposed recently and applied successfully to a simplified mono-dimensional problem. In this paper, this technique is investigated in the case of three-dimensional poroelastic problems. The technique is first recalled, then analytical proof of the stability of the model are given followed by considerations of numerical improvements of the method. An energetic interpretation of the generalized complex modes is then given and some numerical results are presented to illustrate the performance of the approach.     

大雷诺数Navier-Stokes方程的两水平亚格子模型稳定化方法

基于两重网格离散方法,提出三种求解大雷诺数定常Navier-Stokes方程的两水平亚格子模型稳定化有限元算法.其基本思想是首先在一粗网格上求解带有亚格子模型稳定项的Navier-Stokes方程,然后在细网格上分别用三种不同的校正格式求解一个亚格子模型稳定化的线性问题,以校正粗网格解.通过适当的稳定化参数和粗细网格尺寸的选取,这些算法能取得最优渐近收敛阶的有限元解.最后,用数值模拟验证三种算法的有效性.     

Local structure-preserving methods for the generalized Rosenau-RLW-KdV equation with power law nonlinearity

Local structure-preserving algorithms including multi-symplectic, local energy-and momentum-preserving schemes are proposed for the generalized Rosenau–RLW–Kd V equation based on the multi-symplectic Hamiltonian formula of the equation. Each of the present algorithms holds a discrete conservation law in any time–space region. For the original problem subjected to appropriate boundary conditions, these algorithms will be globally conservative. Discrete fast Fourier transform makes a significant improvement to the computational efficiency of schemes. Numerical results show that the proposed algorithms have satisfactory performance in providing an accurate solution and preserving the discrete invariants.     

Spatial evolution character of multi-objective evolutionary algorithm based on self-organized criticality theory

This paper analyzes the spatial evolution character of multi-objective evolutionary algorithms using self-organized criticality theory. The spatial evolution character is modeled by the statistical property of crowding distance, which displays a scale-free feature and a power-law distribution. We propose that the evolutional rule of multi-objective optimization algorithms is a self-organized state transition from an initial scale-free state to a final scale-free state. The target is to get close to a critical state representing the true Pareto-optimal front. Besides, the anti-Matthew effect is the internal incentive factor of most strategies. The final scale-free state reflects the quality of the final Pareto-optimal front. The speed of the state transition reflects the efficiency of the algorithm. We simulate the spatial evolution characters of three typical multi-objective evolutionary algorithms representing three fields, i.e., Genetic Algorithm, Differential Evolution and the Artificial Immune System algorithm. The results prove that the model and the explanation are effective for analyzing the evolutional rule of multi-objective evolutionary algorithms.     

CAD系统图像颜色分割的新方法

通过图像扫描仪获得的彩色图像常包含有一些样稿颜色以外的杂色或噪声颜色,当用基于一维直方图的传统的颜色统计法进行颜色分割时,会出现错分现象。对此,本文提出了基于二维直方图的颜色分割新方法,实现了一些ＣＡＤ系统图像颜色的准确分割和归并。实验结果和实际应用都证明了新方法的有效性,该方法也可应用于全真彩色图像的量化。     

A New RWA Algorithm Based on Multi-Objective

In this article , we studied the associated research problems and challenges on routing and wavelength assignment (RWA) in WDM (wavelength division multiplexing) networks. Various RWA approaches are examined and compared.We proposed a new RWA algorithm based on multi-objective. In this new algorithm, we consider multiple network optimizing objectives to setup a lightpath with maximize profit and shortest path under the limited resources. By comparing and analyzing, the pro posed algorithm is much better than the algorithms, which only consider one optimizing objective.     

一种快速稀疏贝叶斯学习的水声目标方位估计方法研究

针对以具有时序结构的稀疏贝叶斯学习(Temporally multiple sparse Bayesian learning,TMSBL)为重构算法的水声目标DOA (Direction-of-arrival)估计方法存在运算速度慢的问题,结合块稀疏贝叶斯学习(Block-spare Bayesian learning,BSBL)理论框架下DOA估计模型与特点,采用MacKay提出的定点方法(Fixed-point method)对TMSBL算法中的核心超参量进行求解,提出一种快速的水声目标方位估计稀疏贝叶斯学习的方法,该方法具有运算速度快,重构概率高的特点,并通过实验仿真从运算时间、失败率和均方根误差等方面与TMSBL算法进行比较,验证了该方法的可行性与有效性。      

动态背景下基于光流场分析的运动目标检测算法

针对现有动态背景下运动目标检测算法的不足,提出一种基于光流场分析的运动目标检测算法.首先根据前背景在光流梯度幅值和光流矢量方向上的差异确定目标的大致边界,然后通过点在多边形内部原理获得边界内部的稀疏像素点,最后以超像素为节点,利用混合高斯模型拟合的表观信息和超像素的时空邻域关系构建马尔可夫随机场模型的能量函数,并通过使目标函数能量最小化得到最终的运动目标检测结果.该算法不需要任何先验假设,能够同时处理动态背景和静态背景两种情况.多组实验结果表明,本文算法在检测的准确性和处理速度上均优于现有算法.     

Multi-objective optimization of inverse planning for accurate radiotherapy

The multi-objective optimization of inverse planning based on the Pareto solution set, according to the multi-objective character of inverse planning in accurate radiotherapy, was studied in this paper. Firstly, the clinical requirements of a treatment plan were transformed into a multi-objective optimization problem with multiple constraints. Then, the fast and elitist multi-objective Non-dominated Sorting Genetic Algorithm (NSGA-Ⅱ) was introduced to optimize the problem. A clinical example was tested using this method. The results show that an obtained set of non-dominated solutions were uniformly distributed and the corresponding dose distribution of each solution not only approached the expected dose distribution, but also met the dose-volume constraints. It was indicated that the clinical requirements were better satisfied using the method and the planner could select the optimal treatment plan from the non-dominated solution set.     

Real and Pseudo Pedestrian Detection Method with CA-YOLOv5s Based on Stereo Image Fusion

With the development of convolutional neural networks, the effect of pedestrian detection has been greatly improved by deep learning models. However, the presence of pseudo pedestrians will lead to accuracy reduction in pedestrian detection. To solve the problem that the existing pedestrian detection algorithms cannot distinguish pseudo pedestrians from real pedestrians, a real and pseudo pedestrian detection method with CA-YOLOv5s based on stereo image fusion is proposed in this paper. Firstly, the two-view images of the pedestrian are captured by a binocular stereo camera. Then, a proposed CA-YOLOv5s pedestrian detection algorithm is used for the left-view and right-view images, respectively, to detect the respective pedestrian regions. Afterwards, the detected left-view and right-view pedestrian regions are matched to obtain the feature point set, and the 3D spatial coordinates of the feature point set are calculated with Zhengyou Zhang’s calibration method. Finally, the RANSAC plane-fitting algorithm is adopted to extract the 3D features of the feature point set, and the real and pseudo pedestrian detection is achieved by the trained SVM. The proposed real and pseudo pedestrian detection method with CA-YOLOv5s based on stereo image fusion effectively solves the pseudo pedestrian detection problem and efficiently improves the accuracy. Experimental results also show that for the dataset with real and pseudo pedestrians, the proposed method significantly outperforms other existing pedestrian detection algorithms in terms of accuracy and precision.     

Accurate and efficient algorithm for simulation of fringe projection system

Because of intrinsic drawbacks of existing algorithm for optical simulation: first, there is an accuracy problem, finite-precision machine arithmetic introduces error into line-surface intersection calculation. Second, these algorithms do not or partially consider process of shadow modeling. In this paper, we present an algorithm based on geometric analysis to simulate fringe projection system, a variable-step-iterative (VSI) algorithm is used to search line-surface intersection point, which can avoid many difficult problems. The proposed algorithm can accurately identify the shadow. Simulation results shows the reconstructed RMS errors obtained by our algorithm are far smaller than those obtained by previous reported algorithm. Simulation results validate our VSI algorithm can obtain more accurate results than many previous intersection point algorithms. Simulation results indicate our shadow modeling algorithm exceeds many existing shadow modeling algorithm, and is equivalent to shadow volumes algorithm, but our shadow algorithm runs faster than shadow volumes algorithm.     

机器人自适应主动同时定位与建图新方法

针对多机器人的定位与建图受到即时定位与地图构建(SLAM)研究方法和技术不成熟的制约问题,提出一种基于扩展卡尔曼滤波(EKF)的自适应同时定位与建图方法。首先,基于EKF估计方法,将SLAM中机器人运动方式的选取问题转化为一个多目标最优控制问题,机器人选取最优化目标函数的控制输入,从而以主动的、智能的和自适应的方式探索环境;然后,将上述方法推广到多机器人SLAM中,以实现更为准确、高效和鲁棒的定位与建图。仿真结果表明,该方法大大提高了机器人建图的效率、准确性和鲁棒性。该方法用于机器人主动同时定位和建图是可行的、有效的。     

Detectability of infrared small targets

Selecting the most appropriate algorithms for detecting small targets in varied infrared image scenes is frequently needed, since the relative characteristics between small targets and backgrounds in varied scenes are disparate. To solve that problem, a novel criterion is proposed in this article to measure the difficulty in distinguishing small targets from infrared images, which is named detectability of infrared small targets (DIST). Then a new method to detect small targets based on it was introduced. Detection probability and false alarm probability were employed to evaluate the performance. The results from the experiments further demonstrated the robustness of the proposed methodology under different backgrounds.