Huge Solution Space Research Articles

An AI-assisted terahertz reconfigurable metamaterial in standard 180-nm CMOS

For large-scale terahertz (THz) reconfigurable metamaterials (RM), each unit element needs to be controlled by an independent voltage source. This leads to a huge solution space for control words to achieve the required RM performance specification. In this paper, we propose a fast AI-assisted control-word generation scheme to reduce the computation cost of electromagnetic (EM) simulations (Forward Model) and to speed up iterations for control word searches (Inverse Model). We demonstrate the effectiveness of our models by designing and fabricating a 12 × 9 RM array with a Bi-layer Bi-Split-Ring Resonator structure (B-BSRR) in standard 180 nm CMOS technology. The terahertz RM can be electrically reconfigured, providing active metamaterials with an amplitude modulation depth of around 0–1.7 dB at 0.49 THz. The results show that our Forward Model can predict the S-parameters between 100 GHz and 800 GHz with a mean absolute error of 0.61 dB. Our method is more than 200 times faster than traditional full-wave simulation methods. The Inverse Model can generate required control words within a 1.3 dB error margin in less than 200 iterations.

Multi-Adaptive Strategies-Based Higher-Order Quantum Genetic Algorithm for Agile Remote Sensing Satellite Scheduling Problem.

The agile remote sensing satellite scheduling problem (ARSSSP) for large-scale tasks needs to simultaneously address the difficulties of complex constraints and a huge solution space. Taking inspiration from the quantum genetic algorithm (QGA), a multi-adaptive strategies-based higher-order quantum genetic algorithm (MAS-HOQGA) is proposed for solving the agile remote sensing satellites scheduling problem in this paper. In order to adapt to the requirements of engineering applications, this study combines the total task number and the total task priority as the optimization goal of the scheduling scheme. Firstly, we comprehensively considered the time-dependent characteristics of agile remote sensing satellites, attitude maneuverability, energy balance, and data storage constraints and established a satellite scheduling model that integrates multiple constraints. Then, quantum register operators, adaptive evolution operations, and adaptive mutation transfer operations were introduced to ensure global optimization while reducing time consumption. Finally, this paper demonstrated, through computational experiments, that the MAS-HOQGA exhibits high computational efficiency and excellent global optimization ability in the scheduling process of agile remote sensing satellites for large-scale tasks, while effectively avoiding the problem that the traditional QGA has, namely low solution efficiency and the tendency to easily fall into local optima. This method can be considered for application to the engineering practice of agile remote sensing satellite scheduling for large-scale tasks.

Optimizing flexible job shop scheduling with automated guided vehicles using a multi-strategy-driven genetic algorithm

The flexible job scheduling problem with automated guided vehicles (FJSP-AGVs) is a simplified model of some real manufacturing industries. It contains three strongly coupled subproblems: operation sequences assignment, machine selection, and automatic guided vehicle selection, leading to a huge solution space. Its several unresolved challenges, i.e., problem model and algorithmic designing, persist. Therefore, we first adopt the sequence-based modeling method to establish a mixed-integer linear programming model with makespan, and its correctness is verified by using the Gurobi solver. Subsequently, a multi-strategy-driven genetic algorithm (Mult_stra_GA) is proposed based on the implicit features of FJSP-AGVs. In Mult_stra_GA, for the operation sequence (OS) and the machine assignment (MS) subproblems, we design three targeted strategies, i.e., two layer-based encoding and decoding strategy, a multiple heuristics-based initialization strategy, double crossover, and dual mutation operators. Meanwhile, the problem-specific diversity checking and restart strategies are introduced to avoid Mult_stra_GA falling into local optima. Finally, we conduct experiments on four well-known benchmarks. Through the statistical analysis, the outcomes demonstrate that the Mult_stra_GA algorithm exhibits efficacy when contrasted with other advanced algorithms.

Efficient heuristic methods for berth allocation at multi-line, multi-berth curbside bus stops

Transit management agencies often pre-allocate a multi-berth stop’s berths to specific bus lines so that passengers can find the right place to wait for their buses. Developing optimal berth allocation plans that minimize the total bus delay is essential for mitigating bus queues at busy multi-berth stops. However, this problem is challenging due to the huge solution space, the high degree of stochasticity in bus queues, and the resulting extremely high computational cost. In this paper, we first propose a simple heuristic method inspired by queueing theory. It is based on the idea that evenly distributing the total traffic intensity (defined as the total bus arrival rate times the mean dwell time) among all the berths would produce a lower bus delay. Numerical results demonstrate that this simple method generated very good berth allocation plans (with optimality gaps <6% for no-overtaking and free-overtaking stops) in seconds! It does not rely on time-consuming simulation surrogate models or numerous input data such as the stochastic bus arrival processes and dwell time distributions of each bus line. To further improve the simple heuristic’s performance (especially for limited-overtaking stops), we develop a cluster-based nested partition algorithm that can find a near-optimal plan (e.g., with an optimality gap of <3%) in a much shorter time than a previous algorithm. The algorithm employs the simple heuristic plan as the initial solution. Our methods can be applied to stops with various berth numbers, different proximities to nearby traffic signals, and under diverse bus queueing rules.

Learning Physically Realizable Skills for Online Packing of General 3D Shapes

We study the problem of learning online packing skills for irregular 3D shapes , which is arguably the most challenging setting of bin packing problems. The goal is to consecutively move a sequence of 3D objects with arbitrary shapes into a designated container with only partial observations of the object sequence. We take physical realizability into account, involving physics dynamics and constraints of a placement. The packing policy should understand the 3D geometry of the object to be packed and make effective decisions to accommodate it in the container in a physically realizable way. We propose a Reinforcement Learning (RL) pipeline to learn the policy. The complex irregular geometry and imperfect object placement together lead to huge solution space. Direct training in such space is prohibitively data intensive. We instead propose a theoretically provable method for candidate action generation to reduce the action space of RL and the learning burden. A parameterized policy is then learned to select the best placement from the candidates. Equipped with an efficient method of asynchronous RL acceleration and a data preparation process of simulation-ready training sequences, a mature packing policy can be trained in a physics-based environment within 48 hours. Through extensive evaluation on a variety of real-life shape datasets and comparisons with state-of-the-art baselines, we demonstrate that our method outperforms the best-performing baseline on all datasets by at least 12.8% in terms of packing utility. We also release our datasets and source code to support further research in this direction. 1

A Spark-Based Parallel Implementation of Arithmetic Optimization Algorithm

Arithmetic optimization algorithm (AOA) is a recent population-based metaheuristic widely used for solving optimization problems. However, the emerging large-scale optimization problems pose a great challenge for AOA due to its prohibitive computational cost to traverse the huge solution space effectively. This article proposes a parallel Spark-AOA using Scala on Apache Spark computing platform. Spark-AOA leverages the intrinsic parallel nature of the population-based AOA and the native iterative in-memory computation support of Spark through resilient distributed datasets (RDD) to accelerate the optimization process. Spark-AOA divides the solutions population into several subpopulations that are distributed into multiple RDD partitions and manipulated concurrently. Simulation experiments on different benchmark functions with up to 1,000-dimension and three engineering design problems demonstrate that Spark-AOA outperforms considerably standard AOA and Spark-based implementations of two recent metaheuristics both in terms of run-time and solution quality.

Development and research of triangle-filter convolution neural network for fuel reloading optimization of block-type HTGRs

The problem of fuel reloading optimization is very demanding, which requires to search for the optimal suitable core configuration within a very huge solution space. To solve this problem, updating of optimization solutions and quantitative evaluation of core configurations are two main steps. With the development of computing science, the meta-heuristic algorithms (MHAs) turn out to be effective for updating optimization solutions; while the artificial neural networks (ANNs) proves to be practical for the other. However, there is few research for the application of convolution neural networks (CNNs). For any given image-like inputs, CNNs could quickly calculate the outputs mainly by convolution operators, using rectangle filters to scan the full inputs. However, the geometry of some reactors is not rectangle so that using rectangle filters fails to traverse. Aiming at solving the problems of fuel reloading optimization for block-type HTGRs, the triangle-filter convolution neural networks (TFCNNs) are developed. At the same time, uniform method and ring-ranked method are also developed for determining the weights of TFCNNs. Combined with genetic algorithm and particle swarm optimization, two wide-applied MHAs, TFCNNs are applied to solve the problems of fuel reloading optimization for block-type HTGRs. The results show that the TFCNNs could greatly predict the optimization parameters of testing data set with accuracy higher than 90%. In addition, comparing to DRAGON V4 program, the calculation time of TFCNNs is even less than 1.30% of it, while comparing to BP-ANNs, the ratio of feasible solutions by TFCNNs is evidently higher.

Scheduling non-permutation flowshop with finite buffers and two competitive agents

Flowshop scheduling is a popular and valuable optimisation model in academia. Application of the scheduling model for industrial problems has to consider real-world scenarios, such as non-permutation scheduling, finite buffers and individual competition from different customers. These non-negligibly factors can lead to a huge solution space, leaving it barely able to be effectively solved. Therefore, this study investigates a non-permutation bi-agent flowshop scheduling problem with limited buffers, where tasks are released over time and minimisation of a linear criterion (i.e. weighted combination of makespans) improves customer satisfaction. Mixed integer programming model is established for this strong NP-hardness problem that the optimum cannot be found in polynomial time. For small-scale instances, a branch and bound (B&B) algorithm is provided to achieve exact solutions, where effective non-delay pruning rules, well-designed lower and upper bounds are utilised to eliminate invalid nodes during searching. For medium-scale instances, a hybrid particle swarm optimisation (HPSO) algorithm is proposed to seek high-quality solutions, where a two-stage collaborative evolution promotes deep exploration, an efficient sequence-based rule amends infeasible solutions and several validated mechanisms enhances convergence. Evaluation of extensive computational experiments, including components evaluation and nonparametric test, demonstrate the effectiveness of the presented algorithms.

An anchor-free object detector based on soften optimized bi-directional FPN

We propose an anchor-free object detector that combines a weighted bi-directional Feature Pyramid Network (BiFPN) and Soft Anchor Point Detector to address the object detection problem in a pixel-wise paradigm. The current mainstream object detection methods are anchor-based, which require to set hyper parameters such as scale and aspect ratio. This requires strong prior knowledge and can be difficult to design. Therefore, we propose an anchor-free detector that completely avoids the complex calculations and all the hyper parameters related to the anchor box by eliminating the predefined set of anchor boxes in an anchor-free way. Anchor-free detectors are essentially dense prediction methods. Although the huge solution space can yield high recall, simple anchor-free methods tend to return too many false positives, which leads to the problem of semantic ambiguity caused by the high overlap of object centers. Therefore, we propose BiFPN to alleviate the impact of high overlap which also effectively addresses the problems related to multi-scale features. Moreover, in order to utilize the power of feature pyramid better, we tackle the issues with a novel training strategy that involves two soften optimization techniques, i.e., soft-weighted anchor points and soft-selected pyramid levels. This training strategy further re-weights the quality of the detection results to make our detection results more stable.

Inspecting the Solution Space of Genome-Scale Metabolic Models.

Genome-scale metabolic models are frequently used in computational biology. They offer an integrative view on the metabolic network of an organism without the need to know kinetic information in detail. However, the huge solution space which comes with the analysis of genome-scale models by using, e.g., Flux Balance Analysis (FBA) poses a problem, since it is hard to thoroughly investigate and often only an arbitrarily selected individual flux distribution is discussed as an outcome of FBA. Here, we introduce a new approach to inspect the solution space and we compare it with other approaches, namely Flux Variability Analysis (FVA) and CoPE-FBA, using several different genome-scale models of lactic acid bacteria. We examine the extent to which different types of experimental data limit the solution space and how the robustness of the system increases as a result. We find that our new approach to inspect the solution space is a good complementary method that offers additional insights into the variance of biological phenotypes and can help to prevent wrong conclusions in the analysis of FBA results.

Enhancing hierarchical surrogate-assisted evolutionary algorithm for high-dimensional expensive optimization via random projection

By remarkably reducing real fitness evaluations, surrogate-assisted evolutionary algorithms (SAEAs), especially hierarchical SAEAs, have been shown to be effective in solving computationally expensive optimization problems. The success of hierarchical SAEAs mainly profits from the potential benefit of their global surrogate models known as “blessing of uncertainty” and the high accuracy of local models. However, their performance leaves room for improvement on high-dimensional problems since now it is still challenging to build accurate enough local models due to the huge solution space. Directing against this issue, this study proposes a new hierarchical SAEA by training local surrogate models with the help of the random projection technique. Instead of executing training in the original high-dimensional solution space, the new algorithm first randomly projects training samples onto a set of low-dimensional subspaces, then trains a surrogate model in each subspace, and finally achieves evaluations of candidate solutions by averaging the resulting models. Experimental results on seven benchmark functions of 100 and 200 dimensions demonstrate that random projection can significantly improve the accuracy of local surrogate models and the new proposed hierarchical SAEA possesses an obvious edge over state-of-the-art SAEAs.

ML-KELM: A Kernel Extreme Learning Machine Scheme for Multi-Label Classification of Real Time Data Stream in SIoT

Social Internet of Things is the fusion carrier of social network and Internet of Things. In the social Internet of Things, millions of different intelligent objects connect and communicate with each other, and social data emerge rapidly, which puts forward higher requirements for the rapid dissemination of valuable information. As an important means of information retrieval, the research and application of multi-label classification technology in social Internet of Things environment is relatively few. Characterized by multi-labels cognitive learning, fast response, concept drifting and huge solution space, the multi-label learning becomes more complicated under data stream environment. To overcome the problems above, this paper proposed a multi-label algorithm based on kernel extreme learning machine. In the training phase, both Cholesky matrix decomposition inverse method and matrix block method were deployed on the Spark platform to solve the inverse problem of large matrix, thus improving the modeling efficiency. In addition, incremental learning of the model was realized from the two aspects of example increment and class increment. The online sequential extreme learning machine is improved to adjust the network weights and realize incremental updating of input dimension. Meanwhile, the output layer nodes were divided and assembled in the way of “concept group,” which made the single classifier model structure scalable. And the label incremental learning is realized in the output dimension. Experimental results on large-scale datasets and practical datasets under stream environment demonstrate that the proposed method provides an efficient solution to multi-label classification.

Learning Asynchronous Boolean Networks From Single-Cell Data Using Multiobjective Cooperative Genetic Programming.

Recent advances in high-throughput single-cell technologies provide new opportunities for computational modeling of gene regulatory networks (GRNs) with an unprecedented amount of gene expression data. Current studies on the Boolean network (BN) modeling of GRNs mostly depend on bulk time-series data and focus on the synchronous update scheme due to its computational simplicity and tractability. However, such synchrony is a strong and rarely biologically realistic assumption. In this study, we adopt the asynchronous update scheme instead and propose a novel framework called SgpNet to infer asynchronous BNs from single-cell data by formulating it into a multiobjective optimization problem. SgpNet aims to find BNs that can match the asynchronous state transition graph (STG) extracted from single-cell data and retain the sparsity of GRNs. To search the huge solution space efficiently, we encode each Boolean function as a tree in genetic programming and evolve all functions of a network simultaneously via cooperative coevolution. Besides, we develop a regulator preselection strategy in view of GRN sparsity to further enhance learning efficiency. An error threshold estimation heuristic is also proposed to ease tedious parameter tuning. SgpNet is compared with the state-of-the-art method on both synthetic data and experimental single-cell data. Results show that SgpNet achieves comparable inference accuracy, while it has far fewer parameters and eliminates artificial restrictions on the Boolean function structures. Furthermore, SgpNet can potentially scale to large networks via straightforward parallelization on multiple cores.

GGA: A modified genetic algorithm with gradient-based local search for solving constrained optimization problems

In the last few decades, genetic algorithms (GAs) demonstrated to be an effective approach for solving real-world optimization problems. However, it is known that, in presence of a huge solution space and many local optima, GAs cannot guarantee the achievement of global optimality. In this work, in order to make GAs more effective in finding the global optimal solution, we propose a hybrid GA which combines the classical genetic mechanisms with the gradient-descent (GD) technique for local searching and constraints management. The basic idea is to exploit the GD capability in finding local optima to refine search space exploration and to place individuals in areas that are more favorable for achieving convergence. This confers to GAs the capability of escaping from the discovered local optima, by progressively moving towards the global solution. Experimental results on a set of test problems from well-known benchmarks showed that our proposal is competitive with other more complex and notable approaches, in terms of solution precision as well as reduced number of individuals and generations.

A hybrid algorithm combining genetic algorithm and variable neighborhood search for process sequencing optimization of large-size problem

ABSTRACT On the premise of satisfying the process priority relationship, there are many kinds of feasible sequencing schemes. How to obtain the optimal process sequencing meeting the process priority relationship has always been a popular research area in the field of CAPP (Computer-Aided Process Planning). Currently, some achievements have been made in the field of small and medium-size problems. For large-size problems, due to the explosion of solution space, the existing bionic algorithms are easy to fall into local optimum or even non-convergence. In this paper, a hybrid algorithm combining genetic algorithm and variable neighborhood search is proposed to solve the above problems. The basic idea is to decompose the complex and huge solution space into relatively simple multi-neighborhood spaces, and then search in each neighborhood space by genetic algorithm in turn. The global optimal solution is obtained when a solution is the best solution through all neighborhood spaces. Based on this idea, the hybrid algorithm framework and neighborhood construction rules are developed, and the implementation steps of the hybrid algorithm are detailed. Taking a real-world case as the case study, the feasibility and superiority of the proposed hybrid algorithm are demonstrated by algorithm comparison tests.

A Novel Service System for Long-Distance Drone Delivery Using the “Ant Colony+A*” Algorithm

Drones, with the potential to significantly increase the efficiency of the delivery, have received much attention in recent years. Still, there are some bottleneck problems in the application of the long-distance drone delivery, such as the limited flight range and flight safety. Therefore, the article proposes a novel service system, including the battery exchange stations and maintenance checkpoints, to provide long-distance delivery services. Then, with respect to the service system, we construct a drone path programming model, where a special penalty value is proposed as the objective function to simultaneously minimize the path length and number of landing depots for the delivery service. Thereafter, to efficiently find the optimal flight path among huge solution space, we improve the ant colony optimization with the A <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">*</sup> algorithm embedded to avoid the nondirectional searching of ants. Finally, we use a case of Shanghai city to study the feasibility and effectiveness of our approaches, which includes the comparison of our algorithm and the other three heuristics on ten random delivery cases, the verification of the effectiveness of our algorithm on the long-distance delivery service, and a sensitive analysis of the effect of the depot number on the optimal solution.

Multi-material design: architecture and components simultaneous selection

The design of multi-materials requires the determination of several variables: morphology, components and geometric parameters. Previously developed methods help designer making these choices separately because they always lie on a reduction in the number of design variables, as some of them are fixed at the beginning of the study. This paper proposes a method to carry out the simultaneous definition of all these variables in a preliminary design step. The beginning of this work consists in the formalisation of the multi-scale decomposition of the architecture materials. This outline is a guide for the use of databases of materials and architectures, whose combination enables the generation of multi-materials. Within the huge solution space resulting from this approach, a genetic algorithm allows to find optimised materials to fulfil a set of requirements with a limited number of calculations. An analytic example illustrates the efficiency of this method and shows it can provide designers with several propositions of materials.

Large-scale point cloud contour extraction via 3D guided multi-conditional generative adversarial network

As one of the most important features for human perception, contours are widely used in many graphics and mapping applications. However, for large outdoor scale point clouds, contour extraction is considerably challenging due to the huge, unstructured and irregular point space, thus leading to massive failure for existing approaches. In this paper, to generate contours consistent with human perception for outdoor scenes, we propose, for the first time, 3D guided multi-conditional GAN (3D-GMcGAN), a deep neural network based contour extraction network for large scale point clouds. Specifically, two ideas are proposed to enable the network to learn the distributions of labeled samples. First, a parametric space based framework is proposed via a novel similarity measurement of two parametric models. Such a framework significantly compresses the huge point data space, thus making it much easier for the network to “remember” target distribution. Second, to prevent network loss in the huge solution space, a guided learning framework is designed to assist finding the target contour distribution via an initial guidance. To evaluate the effectiveness of the pro-posed network, we open-sourced the first, to our knowledge, dataset for large scale point cloud with contour annotation information. Experimental results demonstrate that 3D-GMcGAN efficiently generates contours for the data with more than ten million points (about several minutes), while avoiding ad hoc stages or parameters. Also, the proposed framework produces minimal outliers and pseudo-contours, as suggested by comparisons with the state-of-the-art approaches.

Geometric Deep Learning for Shape Correspondence in Mass Customization by Three-Dimensional Printing

Abstract Many industries, such as human-centric product manufacturing, are calling for mass customization with personalized products. One key enabler of mass customization is 3D printing, which makes flexible design and manufacturing possible. However, the personalized designs bring challenges for the shape matching and analysis, owing to the high complexity and shape variations. Traditional shape matching methods are limited to spatial alignment and finding a transformation matrix for two shapes, which cannot determine a vertex-to-vertex or feature-to-feature correlation between the two shapes. Hence, such a method cannot measure the deformation of the shape and interested features directly. To measure the deformations widely seen in the mass customization paradigm and address the issues of alignment methods in shape matching, we identify the geometry matching of deformed shapes as a correspondence problem. The problem is challenging due to the huge solution space and nonlinear complexity, which is difficult for conventional optimization methods to solve. According to the observation that the well-established massive databases provide the correspondence results of the treated teeth models, a learning-based method is proposed for the shape correspondence problem. Specifically, a state-of-the-art geometric deep learning method is used to learn the correspondence of a set of collected deformed shapes. Through learning the deformations of the models, the underlying variations of the shapes are extracted and used for finding the vertex-to-vertex mapping among these shapes. We demonstrate the application of the proposed approach in the orthodontics industry, and the experimental results show that the proposed method can predict correspondence fast and accurate, also robust to extreme cases. Furthermore, the proposed method is favorably suitable for deformed shape analysis in mass customization enabled by 3D printing.

Pareto-Optimal Transit Route Planning With Multi-Objective Monte-Carlo Tree Search

Planning ideal transit routes in the complex urban environment can improve the performance and efficiency of public transportation systems effectively. However, finding such routes is computationally difficult due to the huge solution space constituted by billions of possible routes. Considering the limited scalability of exact search methods, heuristic search methods were proposed to boost the efficiency and incorporate flexible constraints. Nevertheless, the existing methods conceal multiple criteria in an objective, and thus evaluating the performance of the generated route becomes challenging due to the lack of comparable alternatives. Inspired by the prior study, we formulate the definition of pareto-optimal transit routes based on multiple criteria. However, extracting these routes remains challenging because: A) the sheer volume of possible transit routes; and B) the sparsity of pareto-optimal routes. We address these challenges by developing an efficient search framework: for challenge A, a random search method is developed based on Monte Carlo tree search where the unproductive solution subspaces are pruned progressively to reduce the search cost; and for challenge B, an estimation method is derived to guide the search process by assessing the value for each solution subspace. The superior effectiveness of our approach in approximating the pareto-optimal transit routes was demonstrated by the comprehensive evaluation based on the real-world data.