Showing papers on "Benchmark (computing) published in 2022"

Improving Dendritic Neuron Model With Dynamic Scale-Free Network-Based Differential Evolution

Abstract: Some recent research reports that a dendritic neuron model (DNM) can achieve better performance than traditional artificial neuron networks (ANNs) on classification, prediction, and other problems when its parameters are well-tuned by a learning algorithm. However, the back-propagation algorithm (BP), as a mostly used learning algorithm, intrinsically suffers from defects of slow convergence and easily dropping into local minima. Therefore, more and more research adopts non-BP learning algorithms to train ANNs. In this paper, a dynamic scale-free network-based differential evolution (DSNDE) is developed by considering the demands of convergent speed and the ability to jump out of local minima. The performance of a DSNDE trained DNM is tested on 14 benchmark datasets and a photovoltaic power forecasting problem. Nine meta-heuristic algorithms are applied into comparison, including the champion of the 2017 IEEE Congress on Evolutionary Computation (CEC2017) benchmark competition effective butterfly optimizer with covariance matrix adapted retreat phase (EBOwithCMAR). The experimental results reveal that DSNDE achieves better performance than its peers.

Dynamic Levy Flight Chimp Optimization

Abstract: Background: The Chimp Optimization Algorithm (ChOA) is a hunting-based model and can be utilized as a set of optimization rules to tackle optimization problems. Due to agents’ insufficient diversity in some complex problems, this algorithm is sometimes exposed to local optima stagnation. Objective: This paper introduces a Dynamic Levy Flight (DLF) technique to smoothly and gradually transit the search agents from the exploration phase to the exploitation phase. Methods: To investigate the efficiency of the DLFChOA, this paper evaluates the performance of DLFChOA on twenty-three standard benchmark functions, twenty challenging functions of CEC-2005, ten suit tests of IEEE CEC06-2019, and twelve real-world optimization problems. The results are compared to benchmark optimization algorithms, including CMA-ES, SHADE, ChOA, HGSO, LGWO and ALEP (as the best benchmark Levy-based algorithms), and eighteen state-of-the-art algorithms (as the winners of the CEC2019, the GECCO2019, and the SEMCCO2019). Result and conclusion: Among forty-three numerical test functions, DLFChOA and CMA-ES gain the first and second rank with thirty and eleven best results. In the 100-digit challenge, jDE100 with a score of 100 provides the best results, followed by DISHchain1e+12, and DLFChOA with a score of 85.68 is ranked fifth among eighteen state-of-the-art algorithms achieved the best score in seven out of ten problems. Finally, DLFChOA and CMA-ES respectively gain the best results in five and four real-world engineering problems.

Modelling and optimization of integrated distributed flow shop scheduling and distribution problems with time windows

Abstract: Production and distribution are two essential activities in supply chain management. Currently, integrated production and distribution problems receive much attention because decision-makers devote to improving the operation efficiency of both stages and try to achieve an optimal solution. This work proposes an integrated distributed production and distribution problem with consideration of time windows, in which a set of jobs (i.e., customer orders) needs to be assigned among factories and the jobs are processed on flow shop environments at their associated factories. Then, the completed jobs are delivered by capacitated vehicles to customers in different regions while satisfying given time windows as much as possible. Accordingly, to optimally solve the proposed problem, a mixed integer programming model with minimizing total weighted earliness and tardiness has been established. For the optimization task, an enhanced brain storm optimization algorithm with some particular strategies is designed to handle the considered problem. To assess the performance of the proposed optimization method, several experiments by adopting a set of benchmark test problems are performed, and state-of-the-art optimizers are chosen for comparisons. The obtained optimization results exhibit that the designed algorithm significantly outperforms its rivals and can be considered as an excellent optimizer for solving the studied problem. Besides, compared with the CPLEX solver, the designed optimizer also performs much better for solving large-size problems.

Hosting capacity of photovoltaic systems in low voltage distribution systems: A benchmark of deterministic and stochastic approaches

Abstract: This paper compares and benchmarks different methods to calculate the hosting capacity of low voltage distribution systems for residential photovoltaic generation. The low voltage distribution system hosting capacity provides insight to the network planner and operator regarding the capability of the distribution system to accommodate new photovoltaic installations. In the literature, many different methods to calculate the hosting capacity of the distribution network exist, but they are demonstrated using different feeders and with different assumptions, making it difficult to compare different approaches. In this paper, a consistent benchmark of existing hosting capacity methods is performed. A publicly available real network of a semi-urban area is used to compare the different approaches in a consistent manner. A reference stochastic hosting capacity method is proposed, which considers all sources of uncertainty as well as operational and probabilistic limits. Hosting capacity definitions, based on both deterministic and stochastic methods, are evaluated based on their performance parameters and benchmarked. The performed analysis allows selecting the most appropriate hosting capacity definition and comparing results amongst systems.

Deep Reinforcement Learning for Load-Balancing Aware Network Control in IoT Edge Systems

Abstract: Load balancing is directly associated with the overall performance of a parallel and distributed computing system. Although the relevant problems in communication and computation have been well studied in data center environments, few works have considered the issues in an Internet of Things (IoT) edge scenario. In fact, processing data in a load balancing way for the latter case is more challenging. The main reason is that, unlike a data center, both the data sources and the network infrastructure in an IoT edge system can be dynamic. Moreover, with different performance requirements from IoT networks and edge servers, it will be hard to characterize the performance model and to perform runtime optimization for the whole system. To tackle this problem, in this work, we propose a load-balancing aware networking approach for efficient data processing in IoT edge systems. Specifically, we introduce an IoT network dynamic clustering solution using the emerging deep reinforcement learning (DRL), which can both fulfill the communication balancing requirements from IoT networks and the computation balancing requirements from edge servers. Moreover, we implement our system with a long short term memory (LSTM) based Dueling Double Deep Q-Learning Network (D3QN) model, and our experiments with real-world datasets collected from an autopilot vehicle demonstrate that our proposed method can achieve significant performance improvement compared to benchmark solutions.

Self-adaptive resources allocation-based differential evolution for constrained evolutionary optimization

Abstract: When using evolutionary algorithms to address constrained optimization problems, it is important to balance not only the diversity and convergence but also the constraints and objective function. To this end, a self-adaptive resources allocation-based differential evolution (SRADE) is presented in this paper. Specifically, during the evolutionary process, three mutation strategies with distinct focuses are collaboratively employed and adaptively assigned to different individuals based on their performance feedback. That is, most of the computing resources will be consumed by the most efficient strategy at different evolutionary stages to mitigate inefficient search under limited resources. These three collaborative strategies focus on maintaining population diversity, driving the population into feasible regions, and promoting the population toward the objective, respectively. Combining the self-adaptive resources allocation scheme and diverse search strategies is expected to satisfy the requirements of the population for diversity, convergence, constraints, and the objective at a certain iteration. Extensive experiments are performed on three benchmark test suites, including a large number of test functions from IEEE CEC 2006, 2010, and 2017. Compared to other well-designed constrained evolutionary approaches, SRADE exhibits superior or very competitive performance.

Semiconductor final testing scheduling using Q-learning based hyper-heuristic

Abstract: Semiconductor final testing scheduling problem (SFTSP) has extensively been studied in advanced manufacturing and intelligent scheduling fields. This paper presents a Q-learning based hyper-heuristic (QHH) algorithm to address the SFTSP with makespan criterion. The structure of QHH employs the Q-learning algorithm as the high-level strategy to autonomously select a heuristic from a pre-designed low-level heuristic set. The selected heuristic in different stages of the optimization process is recognized as the executable action and performed on the solution space for better results. An efficient encoding and decoding pair is presented to generate feasible schedules, and a left-shift scheme is embedded into the decoding process for improving resources utilization. Additionally, the design-of-experiment method is implemented to investigate the effect of parameters setting. Both computational simulation and comparison are finally carried out on a benchmark set and the results demonstrate the effectiveness and efficiency of the proposed QHH.

Modelling and Scheduling Integration of Distributed Production and Distribution Problems via Black Widow Optimization

Abstract: Production and distribution are two important sectors in a supply chain and their managements become an essential issue in industrial fields. The integrated operation of production and distribution stages are regarded as an effective approach. This work proposes an integrated production and distribution optimization problem, where jobs are processed in a distributed manufacturing system with multiple flow shops, and then they are delivered to customers locating in geographically-dispersed points. To mathematically describe this problem, a mixed integer programming model is formulated to minimize maximum completion time. In order to optimally solve the proposed problem, an enhanced black widow optimization algorithm is developed to deal with the studied problem. In this proposed approach, the solution representation, population initialization, procreation, cannibalism, and mutation along with a simulated annealing approach are specially designed. Then, a design of experiment approach is employed to analyze the influence of sensitive parameters on the proposed approach. Performance and efficiency of the designed method are validated through conducting extensive experiments on a set of benchmark test problems. Besides, comparisons with some well-known optimizers in the literature have been conducted to show the superiority of the proposed method.

A multi-objective elitist feedback teaching–learning-based optimization algorithm and its application

Abstract: Students generally attempt to improve themselves with different manners in their spare time. Inspired by this fact, this paper presents a multi-objective elitist feedback teaching–learning-based optimization (MEFTO) algorithm for multi-objective optimization problems. To promote the exploration capacity and convergence of the basic teaching–learning-based optimization (TLBO), a feedback phase that simulates the spare-time learning phenomenon is introduced at the end of the learner phase. Students should compare their score with the average class score to select an appropriate means for further improvement. Poorly performing students can learn from the teacher directly for rapid improvement, whereas high-performing students prefer to motivate themselves for reinforcement learning. Non-dominated sorting is incorporated to permit this heuristic to solve problems with several objectives. Crowding distance calculation is adopted to maintain the diversity of the obtained solution set in a single run. Elitism strategy is also employed to provide a great improvement for the algorithm. The performance of the MEFTO algorithm is compared with seven other well-known algorithms by using a set of unconstrained benchmark test problems and three constrained engineering optimization problems. The qualitative and quantitative results indicate that the proposed algorithm can provide considerably competitive results and outperforms the other algorithms.

Analysis and Compensation of Benchmark Drift of Micromachined Thermal Wind Sensor Caused by Packaging Asymmetry

Abstract: In this article, the benchmark drift of micromachined thermal wind sensor due to the packaging asymmetry has been analyzed and compensated. For the ideal wind sensors with symmetric packaging, the benchmark does not vary with speed and direction. However, for the real sensors, especially for handmade ones, packaging asymmetry problems are usually inevitable. Consequently, benchmark drifts with wind because of different thermal distribution in each direction, which was verified by the proposed lumped parameters model. The real sensors were tested in a wind tunnel and results showed the benchmark drift as well. The contribution of this work lies in the establishment of the equation of benchmark dependent on wind speed, which can predict benchmark value under different wind speeds by calibration at zero wind speed. This compensation method is extremely simple and convenient for industrial production. After compensation, the relative wind speed errors are reduced from 46 to 8% up to 30 m/s, and the wind direction errors are reduced from 30° to 7° over the full range of 360°.

Modified Particle Swarm Optimization with Unique Self-cognitive Learning for Global Optimization Problems

Abstract: Although different modified versions of particle swarm optimization (PSO) were proposed in past decades to solve global optimization problems, the appropriate mechanism used to attain proper balancing of algorithm’s exploration and exploitation searches remains as an open-ended challenges. A modified PSO with unique self-cognitive learning (MPSO-USCL) is proposed in this paper to address this issue. For each particle, a unique exemplar can be generated by the proposed USCL module to replace the self-cognitive component of each particle and guide its search process towards the promising regions of search space with different levels of exploration and exploitation strengths. Extensive simulation studies are performed to compare the optimization performances of MPSO-USCL with six existing PSO variants using 12 benchmark functions. The proposed MPSO-USCL is reported to outperform its peer algorithms for all benchmark functions.

Hybrid Manta Ray Foraging—Particle Swarm Algorithm for PD Control Optimization of an Inverted Pendulum

Abstract: This paper presents a hybrid Manta ray foraging—particle swarm optimization algorithm. Manta Ray Foraging Optimization (MRFO) algorithm is a recent algorithm that has a promising performance as compared to other popular algorithms. On the other hand, Particle Swarm Optimization (PSO) algorithm is a well-known and a good performance algorithm. The proposed hybrid algorithm in this work incorporates social interaction and elitism mechanisms from PSO into MRFO strategy. The mechanisms help search agents to determine their new search direction. The proposed algorithm is tested on various dimensions and fitness landscapes of CEC2014 benchmark functions. In solving a real world engineering problem, it is applied to optimize a PD controller for an inverted pendulum system. Result of the benchmark function test is statistically analyzed. The proposed algorithm has successfully improved the accuracy performance for most of the test functions. For optimization of the PD control, result shows that the proposed algorithm has attained a better control performance compared to MRFO.

Exploring more diverse network architectures for single image super-resolution

Abstract: We propose a plug-and-play neural architecture search (NAS) method to explore diverse architectures for single image super-resolution (SISR). Unlike current NAS-based methods with the single path setting and pipeline setting, our proposed method achieves the trade-off between diverse network architectures and search cost. Our proposed method formulates the task in a differentiable manner, which inherits the architecture parameter optimization method from Discrete Stochastic Neural Architecture Search (DSNAS). Besides the straightforward searching of operations, we also search each node in a cell for the activation function, from-node, and skip-connection node, which diverse the searched architecture topologies. The individually searching of skip-connection node avoids skip-connection excessive phenomenon. Moreover, to alleviate the influence of inconsistent architecture between training and testing periods, we introduce random variables into the architecture parameter as regularization. Benchmark experiments show our state-of-the-art performance under specific parameters and FLOPs constraints. Compared with other NAS-based SISR methods, our proposed methods achieve better performance with less searching time and resources. The superior results further demonstrate the effectiveness of our proposed NAS methods.

Deep Learning-Based Inverse Modeling for Predictive Control

Abstract: Deep learning architectures can accurately model complex dynamics such as Inverse models of nonlinear systems. Their uses in the control context have been well-developed in the past forming a strong basis. However, the black-box nature of such techniques have limited its usability in real- world applications. This work aims to provide an auxiliary framework that combines intelligent data-based information with the conventional Predictive Control approach. By adding the inverse model signals to the closed-loop optimization search, the performance and accuracy are improved in our test cases. We provide a set of benchmark experiments as well as different optimization algorithms that benefit from Inverse models. Our results show a clear addition of Intelligent mechanisms to the standard methodology, without the uncertainties of black-box models and with all the advantages.

TaintBench: Automatic real-world malware benchmarking of Android taint analyses

Abstract: Due to the lack of established real-world benchmark suites for static taint analyses of Android applications, evaluations of these analyses are often restricted and hard to compare. Even in evaluations that do use real-world apps, details about the ground truth in those apps are rarely documented, which makes it difficult to compare and reproduce the results. To push Android taint analysis research forward, this paper thus recommends criteria for constructing real-world benchmark suites for this specific domain, and presents TaintBench, the first real-world malware benchmark suite with documented taint flows. TaintBench benchmark apps include taint flows with complex structures, and addresses static challenges that are commonly agreed on by the community. Together with the TaintBench suite, we introduce the TaintBench framework, whose goal is to simplify real-world benchmarking of Android taint analyses. First, a usability test shows that the framework improves experts’ performance and perceived usability when documenting and inspecting taint flows. Second, experiments using TaintBench reveal new insights for the taint analysis tools Amandroid and FlowDroid: (i) They are less effective on real-world malware apps than on synthetic benchmark apps. (ii) Predefined lists of sources and sinks heavily impact the tools’ accuracy. (iii) Surprisingly, up-to-date versions of both tools are less accurate than their predecessors.

Performance Analysis of Genetic Algorithm for Function Optimization in Multicore Platform Using DEAP

Abstract: Meta-heuristic algorithms are applied to find good or near-optimal solutions at a reasonable computational cost and time by exploring the search space in an efficient way. Parallelization and distributed computing techniques are solutions to enhance the algorithmic performance. The objective of this work is to analyze the performance of genetic algorithm for function optimization in multicore platform using distributed evolutionary algorithms in Python (DEAP) framework. The analysis is done based on optimal value obtained and the execution time taken to run benchmark functions. Ten benchmark functions of fixed dimensions and five benchmark functions with variable dimensions are considered during experimentation. From the results, we infer that the performance of GA in multicore platform is efficient in terms of computing speed and finding optimal value for the selected standard benchmark problems.