Optimization-based Search Research Articles

Flexi-BOPI: Flexible granularity pipeline inference with Bayesian optimization for deep learning models on HMPSoC

To achieve high-throughput deep learning (DL) model inference on heterogeneous multiprocessor systems-on-chip (HMPSoC) platforms, the use of pipelining for the simultaneous utilization of multiple resources has emerged as a promising solution. Nevertheless, current research faces two primary challenges: determining the optimal pipeline partitioning granularity, which directly influences the inference performance, and addressing the high time overhead of the search algorithms. To address these challenges, we propose Flexi-BOPI, a pipeline inference method for DL models of HMPSoCs. Flexi-BOPI offers flexible pipeline partitioning granularity down to a minimum size of a single core, enhancing the performance by better adapting to the diverse computational demands of different layers in DL models. Flexi-BOPI employs a Bayesian optimization-based search algorithm to significantly reduce the search overhead. In addition, we propose a surrogate model based on the heteroscedastic Gaussian process (HGP) to address the challenge of sample noise during the evaluation process. This approach can further reduce search overhead. Our experimental results demonstrate that the proposed method achieves significant improvements in inference performance and search overhead compared to existing methods.

Specialized path-based technique to test Internet of Things system functionality under limited network connectivity

Contemporary Internet-of-Things (IoT) systems are hindered by several reliability-related issues, especially, the dynamic behavior of IoT systems caused by limited and often unstable network connectivity. Several intuitive ad-hoc approaches can be employed to test this behavior; however, the effectiveness of these approaches in detecting defects and their overall testing costs remain questionable. Therefore, we present a new specialized path-based technique to test the processes of an IoT system in scenarios wherein parts of these processes are influenced by limited or disrupted network connectivity. The proposed technique can be scaled using two levels of test coverage criteria to determine the strengths of the test cases. For this purpose, we propose two algorithms for generating test cases to implement the technique: an ant colony optimization-based search and a graph-traversal-based test case composition. We compared the efficiency of the proposed approach with possible solutions obtained using a standard path-based testing approach based on prime paths computed by a set-covering algorithm. We consider the total number of test case steps as the main proxy for test effort in experiments employing 150 problem models. For the less intensive of the two used test-coverage criteria, EachBorderOnce, an ant colony optimization-based algorithm, produced test sets with the same averaged number of steps as the graph traversal-based test-case composition; however, this algorithm performed with averaged number of steps 10% lower than a prime paths-based algorithm. For the more intensive test coverage criterion, AllBorderCombinations, these differences favoring the ant colony optimization-based algorithm were 18% and 25%, respectively. For these two types of defined test coverage criteria, the ant colony optimization-based search, graph-traversal-based algorithm, and standard path-based testing approach based on prime paths achieved the best results for 93 and 78, 14 and 24, and 13 and 17 models for AllBorderCombinations and EachBorderOnce criterion, respectively. Therefore, to guarantee the best test set, all compared algorithms are combined in a portfolio strategy that yields the best results based on the potential of the produced test sets to detect simulated defects caused by limited network connectivity. Additionally, this portfolio strategy also yields test sets, implying the lowest test effort for experimental problem instances.

UCAV Path Planning for Avoiding Obstacles using Cooperative Co-evolution Spider Monkey Optimization

The unmanned combat aerial vehicle (UCAV) path planning problem is a complex global optimization problem that aims to seek an optimal flight route that avoids the threats and constraints on the battlefield. However, as the number and degree of the path threats increases, the recent UCAV path planning algorithms all suffer from being trapped into local optima with a low convergence rate, which leads to the poor performance. Therefore, to resolve these problems, we propose a Cooperative co-Evolution (CE)-based Spider Monkey Optimization (SMO) algorithm, named CESMO to address UCAV path planning problem for avoiding obstacles. First, a cooperative co-evolution strategy is proposed to prevent the algorithms from falling into local optima easily, enhancing the coherence of the path. After that, a further division is applied to the algorithm for diversifying the search direction of each individual in the sub-groups. Finally, the spider monkey optimization-based search mechanism is designed to further strengthen the search efficiency and boost the convergence rate. Experimental results demonstrate that our proposed method is more competitive than other state-of-the-art evolutionary algorithms for UCAV path planning problem considering the quality and stability of the final paths.

A swarm optimization-based search algorithm for the quadratic knapsack problem with conflict Graphs

The knapsack problem arises in a variety of real world applications such as railway stations, flexible manufacturing systems, multimedia, cryptography and hydrological studies. In this paper, a special case of the knapsack problem is tackled: the quadratic knapsack problem with conflict graphs. This problem is solved by using a population-based search algorithm, which is inspired from the binary particle swarm optimization combined with a quick and efficient local search. The particle swarm optimization generates a population of particles while the local search procedure tries either to repair the infeasibility of each binary solution or to improve its quality. The performance of the proposed method is evaluated on a set of benchmark instances taken from the literature (containing medium and large-scale instances), where its achieved results are compared to those published in the literature containing the bounds realized with GLPK, Cplex and those achieved by more recent methods. The proposed method remains competitive, where encouraging results have been obtained.

Special Issue on Augmented Prototyping and Fabrication for Advanced Product Design and Manufacturing

“Don’t automate, augment!” This is the takeaway of the seminal book on the future of work by Davenport and Kirby.*1 The emergence of cyber-physical systems makes radical new products and systems possible and challenges the role of humankind. Throughout the design, manufacturing, use, maintenance, and end-of-life stages, digital aspects (sensing, inferencing, connecting) influence the physical (digital fabrication, robotics) and vice versa. A key takeaway is that such innovations can augment human capabilities to extend our mental and physical skills with computational and robotic support – a notion called “augmented well-being.” Furthermore, agile development methods, complemented by mixed-reality systems and 3D-printing systems, enable us to create and adapt such systems on the fly, with almost instant turnaround times. Following this line of thought, our special issue is entitled “Augmented Prototyping and Fabrication for Advanced Product Design and Manufacturing.” Heavily inspired by the framework of Prof. Jun Rekimoto’s Augmented Human framework,*2 we can discern two orthogonal axes: cognitive versus physical and reflective versus active. As depicted in Fig. 1, this creates four different quadrants with important scientific domains that need to be juxtaposed. The contributions in this special issue are valuable steps towards this concept and are briefly discussed below. AR/VR To drive AR to the next level, robust tracking and tracing techniques are essential. The paper by Sumiyoshi et al. presents a new algorithm for object recognition and pose estimation in a strongly cluttered environment. As an example of how AR/VR can reshape human skills training, the development report of Komizunai et al. demonstrates an endotracheal suctioning simulator that establishes an optimized, spatial display with projector-based AR. Robotics/Cyborg Shor et al. present an augmentation display that uses haptics to go beyond the visual senses. The display has all the elements of a robotic system and is directly coupled to the human hand. In a completely different way, the article by Mitani et al. presents a development in soft robotics: a tongue simulator development (smart sensing and production of soft material), with a detailed account of the production and the technical performance. Finally, to consider novel human-robot interaction, human body tracking is essential. The system presented by Maruyama et al. introduces human motion capture based on IME, in this case the motion of cycling. Co-making Augmented well-being has to consider human-centered design and new collaborative environments where the stakeholders involved in whole product life-cycle work together to deliver better solutions. Inoue et al. propose a generalized decision-making scheme for universal design which considers anthropometric diversity. In the paper by Tanaka et al., paper inspection documents are electronically superimposed on 3D design models to enable design-inspection collaboration and more reliable maintenance activities for large-scale infrastructures. Artificial Intelligence Nakamura et al. propose an optimization-based search for interference-free paths and the poses of equipment in cluttered indoor environments, captured by interactive RGBD scans. AR-based guidance is provided to the user. Finally, the editors would like to express their gratitude to the authors for their exceptional contributions and to the anonymous reviewers for their devoted work. We expect that this special issue will encourage a new departure for research on augmented prototyping for product design and manufacturing. *1 T. H. Davenport and J. Kirby, “Only Humans Need Apply: Winners and Losers in the Age of Smart Machines,” Harper Business, 2016. *2 https://lab.rekimoto.org/about/ [Accessed June 21, 2019]

A Computational Design Framework for Efficient, Fabrication Error-Tolerant, Planar THz Diffractive Optical Elements

We demonstrate ultra-thin (1.5-3λ0), fabrication-error tolerant efficient diffractive terahertz (THz) optical elements designed using a computer-aided optimization-based search algorithm. The basic operation of these components is modeled using scalar diffraction of electromagnetic waves through a pixelated multi-level 3D-printed polymer structure. Through the proposed design framework, we demonstrate the design of various ultrathin planar THz optical elements, namely (i) a high Numerical Aperture (N.A.), broadband aberration rectified spherical lens (0.1 THz–0.3 THz), (ii) a spectral splitter (0.3 THz–0.6 THz) and (iii) an on-axis broadband transmissive hologram (0.3 THz–0.5 THz). Such an all-dielectric computational design-based approach is advantageous against metallic or dielectric metasurfaces from the perspective that it incorporates all the inherent structural advantages associated with a scalar diffraction based approach, such as (i) ease of modeling, (ii) substrate-less facile manufacturing, (iii) planar geometry, (iv) high efficiency along with (v) broadband operation, (vi) area scalability and (vii) fabrication error-tolerance. With scalability and error tolerance being two major bottlenecks of previous design strategies. This work is therefore, a significant step towards the design of THz optical elements by bridging the gap between structural and computational design i.e. through a hybrid design-based approach enabling considerably less computational resources than the previous state of the art. Furthermore, the approach used herein can be expanded to a myriad of optical elements at any wavelength regime.

Droplet property optimization in printable electronics fabrication using root system growth algorithm

For printable electronics fabrication, the printing quality delivered by a Drop-on-Demand (DoD) Piezoelectric Inkjet (PIJ) Printhead is limited due to the residual vibration problem in the ink channel. The residual vibration after jetting the ink drop influences the droplet velocity and volume consistency, and limits the maximum jetting frequency of the printhead. In order to meet the challenging requirements of printable electronics fabrication, this work proposes a novel Optimization framework for optimal parameter setting of the high-frequency driving waveform. A novel Root System Growth Algorithm (RSGA), based on principles from plant root growth and foraging behaviors, is chosen as the core optimization algorithm of the framework. The proposed RSGA is benchmarked against other four state-of-the-art bio-inspired algorithms using CEC2005 test function suite. Then several groups of experiment results for various targets are presented to demonstrate the universality of the proposed optimization-based search system for the improvement of printing quality. Simulation results show that the RSGA has an outstanding performance in searching the driving waveform parameters for specified droplet properties.

Optimization of Stimulation Parameters for Targeted Activation of Multiple Neurons Using Closed-Loop Search Methods.

Differential activation of neuronal populations can improve the efficacy of clinical devices such as sensory or cortical prostheses. Improving stimulus specificity will facilitate targeted neuronal activation to convey biologically realistic percepts. In order to deliver more complex stimuli to a neuronal population, stimulus optimization techniques must be developed that will enable a single electrode to activate subpopulations of neurons. However, determining the stimulus needed to evoke targeted neuronal activity is challenging. To find the most selective waveform for a particular population, we apply an optimization-based search routine, Powell’s conjugate direction method, to systematically search the stimulus waveform space. This routine utilizes a 1-D sigmoid activation model and a 2-D strength–duration curve to measure neuronal activation throughout the stimulus waveform space. We implement our search routine in both an experimental study and a simulation study to characterize potential stimulus-evoked populations and the associated selective stimulus waveform spaces. We found that for a population of five neurons, seven distinct sub-populations could be activated. The stimulus waveform space and evoked neuronal activation curves vary with each new combination of neuronal culture and electrode array, resulting in a unique selectivity space. The method presented here can be used to efficiently uncover the selectivity space, focusing experiments in regions with the desired activation pattern.

HiSeeker: Detecting High-Order SNP Interactions Based on Pairwise SNP Combinations.

Detecting single nucleotide polymorphisms’ (SNPs) interaction is one of the most popular approaches for explaining the missing heritability of common complex diseases in genome-wide association studies. Many methods have been proposed for SNP interaction detection, but most of them only focus on pairwise interactions and ignore high-order ones, which may also contribute to complex traits. Existing methods for high-order interaction detection can hardly handle genome-wide data and suffer from low detection power, due to the exponential growth of search space. In this paper, we proposed a flexible two-stage approach (called HiSeeker) to detect high-order interactions. In the screening stage, HiSeeker employs the chi-squared test and logistic regression model to efficiently obtain candidate pairwise combinations, which have intermediate or significant associations with the phenotype for interaction detection. In the search stage, two different strategies (exhaustive search and ant colony optimization-based search) are utilized to detect high-order interactions from candidate combinations. The experimental results on simulated datasets demonstrate that HiSeeker can more efficiently and effectively detect high-order interactions than related representative algorithms. On two real case-control datasets, HiSeeker also detects several significant high-order interactions, whose individual SNPs and pairwise interactions have no strong main effects or pairwise interaction effects, and these high-order interactions can hardly be identified by related algorithms.

Automata Theory Meets Barrier Certificates: Temporal Logic Verification of Nonlinear Systems

We consider temporal logic verification of (possibly nonlinear) dynamical systems evolving over continuous state spaces. Our approach combines automata-based verification and the use of so-called barrier certificates. Automata-based verification allows the decomposition the verification task into a finite collection of simpler constraints over the continuous state space. The satisfaction of these constraints in turn can be (potentially conservatively) proved by appropriately constructed barrier certificates. As a result, our approach, together with optimization-based search for barrier certificates, allows computational verification of dynamical systems against temporal logic properties while avoiding explicit abstractions of the dynamics as commonly done in literature.

Multi-level image thresholding using Otsu and chaotic bat algorithm

Multi-level thresholding is a helpful tool for several image segmentation applications. Evaluating the optimal thresholds can be applied using a widely adopted extensive scheme called Otsu's thresholding. In the current work, bi-level and multi-level threshold procedures are proposed based on their histogram using Otsu's between-class variance and a novel chaotic bat algorithm (CBA). Maximization of between-class variance function in Otsu technique is used as the objective function to obtain the optimum thresholds for the considered grayscale images. The proposed procedure is applied on a standard test images set of sizes (512 × 512) and (481 × 321). Further, the proposed approach performance is compared with heuristic procedures, such as particle swarm optimization, bacterial foraging optimization, firefly algorithm and bat algorithm. The evaluation assessment between the proposed and existing algorithms is conceded using evaluation metrics, namely root-mean-square error, peak signal to noise ratio, structural similarity index, objective function, and CPU time/iteration number of the optimization-based search. The results established that the proposed CBA provided better outcome for maximum number cases compared to its alternatives. Therefore, it can be applied in complex image processing such as automatic target recognition.

Exploring the Effectiveness of Using Graveyard Data When Generating Design Alternatives

The objective of this paper is to demonstrate that unique alternative designs can be efficiently found by searching the discarded data (or graveyard) from a multiobjective genetic algorithm (MOGA). Motivation for using graveyard data to generate design alternatives arises from the computational cost associated with real-time design space exploration of multiobjective optimization problems. The effectiveness of this approach is explored by comparing (1) the uniqueness of alternatives found using graveyard data and those generated using an optimization-based search, and (2) how alternative generation near the Pareto frontier is impacted. Two multiobjective case study problems are introduced—a two bar truss and an I-beam design optimization. Results from these studies indicate that using graveyard data allows for the discovery of alternative designs that are at least 70% as unique as alternatives found using an optimization-based alternative identification approach, while saving a significant number of functional evaluations. Additionally, graveyard data are shown to be better suited for alternative generation near the Pareto frontier than standard sampling techniques. Finally, areas of future work are also discussed.

Reconfiguration of Power Distribution Systems Considering Reliability and Power Loss

A power distribution system reconfiguration methodology considering the reliability and the power loss is developed in this paper. Probabilistic reliability models are used in order to evaluate the reliability at the load points. An algorithm for finding the minimal cut sets is utilized to find the minimal set of components appearing between the feeder and any particular load point. The optimal status of the switches in order to maximize the reliability and minimize the real power loss is found by a binary particle swarm optimization-based search algorithm. The effectiveness of the proposed methodology is demonstrated on a 33-bus and a 123-bus radial distribution system.

Asynchronous particle swarm optimization-based search with a multi-robot system: simulation and implementation on a real robotic system

In this article we consider a version of the Particle Swarm Optimization (PSO) algorithm which is appropriate for search tasks of multi-agent systems consisting of small robots with limited sensing capability. The proposed method adopts asynchronous mechanism for information exchange and position (way point) updates of the agents. Moreover, at each (information exchange) step the agents communicate with only a possibly different subset of the other agents leading to a dynamic neighborhood topology. We implement the algorithm using the Player/Stage realistic robot simulator as well as on real KheperaIII robots using experimentally collected realistic data of ethanol gas concentration. Simulation and implementation results show that the algorithm performs well in a sense that the robots are able to move towards and aggregate in areas with high gas concentration around the maximum points of the gas concentration profile representing the environment.

Decision Support Tool for Life-Cycle-Based Solid Waste Management

Existing solid waste management (SWM) planning software provides only limited assistance to decision makers struggling to find strategies that address their multifarious concerns. The combinatorial nature (many waste items and many management options) and multiple objectives of the SWM problem severely constrain the effectiveness of a manual search process using these tools. Recognizing this, researchers have proposed several optimization-based search procedures. These methods, however, enjoy limited use due to the substantial expertise required for their application. This paper presents a new computer-based decision support framework that addresses these limitations. The new framework integrates process models that quantify the life-cycle inventory of a range of pollutants and costs for an extensive municipal solid waste system, an optimization search procedure that identifies strategies that meet cost and environmental objectives and site-specific restrictions, and a user-friendly interface that facilitates utilization of these components by practitioners. After describing the software design, the use and value of the tool in typical waste management scenarios is demonstrated through a hypothetical, but realistic, case study in which several alternative SWM strategies are generated and examined.