General-purpose Algorithm Research Articles

A Novel Hybrid Harris Hawks Optimization for Color Image Multilevel Thresholding Segmentation

Multilevel thresholding has got more attention in recent years with various successful applications. However, the implementation becomes more and more complex and time-consuming when the number of thresholds is high, and color images which contain more information are even worse. Therefore, this paper proposes an alternative hybrid algorithm for color image segmentation, the advantages of which lie in extracting the best features from the high performance of two algorithms and overcoming the limitations of each algorithm to some extent. Two techniques, Otsu's method, and Kapur's entropy, are used as fitness function to determine the segmentation threshold values. Harris hawks optimization (HHO) is a novel and general-purpose algorithm, and the hybridization of HHO is fulfilled by adding another powerful algorithm-differential evolution (DE), which is known as HHO-DE. More specifically, the whole population is divided into two equal subpopulations which will be assigned to HHO and DE algorithms, respectively. Then both algorithms operate in parallel to update the positions of each subpopulation during the iterative process. In order to fully demonstrate the superior performance of HHO-DE, the proposed method is compared with the seven state-of-the-art algorithms by an array of experiments on ten benchmark images. Meanwhile, five measures, including the average fitness values, standard deviation (STD), peak signal to noise ratio (PSNR), structure similarity index (SSIM), and feature similarity index (FSIM), are used to evaluate the performance of each algorithm. In addition, Wilcoxon's rank sum test for statistical analysis and the comparison with the super-pixel method are also conducted to verify the superiority of HHO-DE. The experimental results reveal that the proposed method significantly outperforms other algorithms. Hence, the HHO-DE algorithm is a remarkable and promising tool for multilevel thresholding color image segmentation.

Deformable soil with adaptive level of detail for tracked and wheeled vehicles

This paper describes a model for deformable soil based on triangular meshes with vertical deformation. Such soil model can be used in a multi-body simulation environment to study the performance of wheeled or tracked vehicles. The formulation is inspired by the soil contact model (SCM), but unlike the original idea, our implementation uses triangular meshes with arbitrary topology. We leverage on this representation of the soil in order to provide a system that automatically refines the level of detail (LOD) of the mesh when tyres or track shoes sink into the soil; this allows the use of coarse meshes as initial approximations of large areas, for the sake of faster performance and lowoverhead on computer memory. A general-purpose collision detection algorithm is used to detect the shape of contacting objects, hence allowing the use of generic geometries to represent lugs, tyre threads, or even track shoes or robotic legs. The method has also been tested with deformable tyres.

Multitask convolutional neural network for no-reference image quality assessment

We propose a multitask convolutional neural network (CNN) for general no-reference image quality assessment (NR-IQA). We decompose the task of rating image quality into two subtasks, namely distortion identification and distortion-level estimation, and then combine the results of the two subtasks to obtain a final image quality score. Unlike conventional multitask convolutional networks, wherein only the early layers are shared and the subsequent layers are different for each subtask, our model shares almost all the layers by integrating a dictionary into the CNN. Moreover, it is trained in an end-to-end manner, and all the parameters, including the weights of the convolutional layers and the codewords of the dictionary, are simultaneously learned from the loss function. We test our method on widely used image quality databases and show that its performance is comparable with those of state-of-the-art general-purpose NR-IQA algorithms.

Dynamic origin-destination matrix calibration for large-scale network simulators

This paper considers offline dynamic OD (origin-destination) calibration problems for large-scale simulation-based network models. We formulate the problem as a simulation-based optimization (SO) problem and propose a scalable and efficient metamodel SO algorithm. For a network with n links and a problem with T time intervals and Z OD pairs per time interval, the corresponding SO problem is a high-dimensional problem of dimension T·Z. At each iteration of the SO algorithm, we solve a set of T independent analytical and differentiable metamodel optimization problems, each of dimension Z. The T analytical problems are constrained by n nonlinear equality constraints. Hence, they scale linearly with the number of links in the network and independently of other network attributes, such as the dimension of the route choice set or the link lengths. The temporal correlation, of the link performance metrics across time intervals, is approximately captured through the parameters of the metamodel. Since the T analytical optimization problems are decoupled and can be solved independently, the proposed approach scales independently of the number of calibration time intervals, making it suitable for the calibration of demand over numerous time periods.The approach is efficient and scalable. It is suitable to address high-dimensional calibration problems for large-scale network models. It is benchmarked on both a toy network and a Singapore network versus two general-purpose algorithms: SPSA and a derivative-free pattern search algorithm. The validation experiments indicate that the proposed method identifies points with objective function estimates that outperform the benchmark methods by 1 to 2 orders of magnitude. As the problem dimension and the temporal correlation, across time intervals, increase, so does the magnitude by which the proposed method outperforms the benchmark methods. The Singapore case study considers a problem of dimension 16,200. This is 2 orders of magnitude higher than what is currently considered high-dimensional for continuous SO problems. The proposed method identifies solutions with an estimated objective function that is 1 order of magnitude better than those of the benchmark methods. It yields solutions that provide an average, across time intervals, improvement to link counts of 77%, compared to the benchmark methods. The case study also illustrates the robustness of the method to the quality of the initial points.

Scheduling Algorithms Using Metaheuristics for Production Processes with Crane Interference

This paper proposes scheduling algorithms using metaheuristics for production processes in which cranes can interfere with each other. There are many production processes that involve cranes in manufacturing industry, such as in the steel industry, so a general purpose algorithm for this problem can be of practical use. The scheduling problem for this process is very complicated and difficult to solve because the cranes must avoid interfering with each other plus each machine has its own operational constraints. Although several algorithms have been proposed for a specific problem or small-scale problem, general purpose algorithms that can be solved in real time (about 30 minutes or less) in the company’s production planning work have not been developed for large-scale problems. This paper develops some metaheuristic algorithms to obtain suboptimal solutions in a short time, and it confirms their effectiveness through computer experiments.

Deep supervised dictionary learning for no-reference image quality assessment

We propose a deep convolutional neural network (CNN) for general no-reference image quality assessment (NR-IQA), i.e., accurate prediction of image quality without a reference image. The proposed model consists of three components such as a local feature extractor that is a fully CNN, an encoding module with an inherent dictionary that aggregates local features to output a fixed-length global quality-aware image representation, and a regression module that maps the representation to an image quality score. Our model can be trained in an end-to-end manner, and all of the parameters, including the weights of the convolutional layers, the dictionary, and the regression weights, are simultaneously learned from the loss function. In addition, the model can predict quality scores for input images of arbitrary sizes in a single step. We tested our method on commonly used image quality databases and showed that its performance is comparable with that of state-of-the-art general-purpose NR-IQA algorithms.

Performance assessment of a multi-objective parametric optimization algorithm with application to a multi-physical engineering system

Engineers routinely perform parameter studies to investigate how system response changes over a range of parameter values. This can provide engineering insight into the importance of an exogenous variable, the robustness of a design alternative, or even the validity of simulation models. Parametric optimization is an extension of this concept in which one investigates how the solution to an optimization or equilibrium problem changes over a range of parameter values. It has been applied in economics, process engineering and engineering systems design. Recent work has yielded a general-purpose algorithm for parametric multi-objective optimization, the Predicted Parametric Pareto Genetic Algorithm (P3GA), and demonstrated it on engineering examples. This article advances understanding about the capabilities of P3GA through a suite of test problems of varying scale and application to a multiphysical engineered system: a magnetohydrodynamic thermal transport (MTT) system. We also compare P3GA to iterated application of an established multi-objective optimization algorithm, NSGA-II. Results indicate P3GA outperforms the iterative approach, underscoring the importance of using an algorithm tailored to parametric optimization.

On Randomness of Compressed Data Using Non-parametric Randomness Tests

Four randomness tests were used to test the outputs (compressed files) of four lossless compressions algorithms: JPEG-LS and JPEG-2000 algorithms are image-dedicated algorithms, while 7z and Bzip2 algorithms are general-purpose algorithms. The relationship between the result of randomness tests and the compression ratio was investigated. This paper reports the important relationship between the statistical information behind these tests and the compression ratio. It shows that, this statistical information almost the same at least, for the four lossless algorithms under test. This information shows that 50 % of the compressed data are grouping of runs, 50% of it has positive signs when comparing adjacent values, 66% of the files containing turning points, and using Cox-Stuart test, 25% of the file give positive signs, which reflects the similarity aspects of compressed data. When it comes to the relationship between the compression ratio and these statistical information, the paper shows also, that, the greater values of these statistical numbers, the greater compression ratio we get.

A generic heuristic for multi-project scheduling problems with global and local resource constraints (RCMPSP)

This paper presents a novel algorithm to solve the multi-project scheduling problem with resource constraints (RCMPSP). The algorithm was tested with all the problems proposed in the multi-project scheduling problem library, which is the main reference to benchmark RCMPSP algorithms. Our analysis of the results demonstrates that this algorithm, in spite of its simplicity, outperforms other algorithms published in the library in 16% of the cases and holds the best result in 27% of the cases. These results, along with the fact that this is a general-purpose algorithm, make it a good choice to deal with limited time and resources in portfolio management.

Group Regularized Estimation Under Structural Hierarchy

ABSTRACTVariable selection for models including interactions between explanatory variables often needs to obey certain hierarchical constraints. Weak or strong structural hierarchy requires that the existence of an interaction term implies at least one or both associated main effects to be present in the model. Lately, this problem has attracted a lot of attention, but existing computational algorithms converge slow even with a moderate number of predictors. Moreover, in contrast to the rich literature on ordinary variable selection, there is a lack of statistical theory to show reasonably low error rates of hierarchical variable selection. This work investigates a new class of estimators that make use of multiple group penalties to capture structural parsimony. We show that the proposed estimators enjoy sharp rate oracle inequalities, and give the minimax lower bounds in strong and weak hierarchical variable selection. A general-purpose algorithm is developed with guaranteed convergence and global optimality. Simulations and real data experiments demonstrate the efficiency and efficacy of the proposed approach. Supplementary materials for this article are available online.

A Family of Scheduling Algorithms for Hybrid Parallel Platforms

More and more parallel computing platforms are built upon hybrid architectures combining multi-core processors (CPUs) and hardware accelerators like General Purpose Graphics Processing Units (GPGPUs). We present in this paper a new method for scheduling efficiently parallel applications with [Formula: see text] CPUs and [Formula: see text] GPGPUs, where each task of the application can be processed either on an usual core (CPU) or on a GPGPU. We consider the problem of scheduling [Formula: see text] independent tasks with the objective to minimize the time for completing the whole application (makespan). This problem is NP-hard, thus, we present two families of approximation algorithms that can achieve approximation ratios of [Formula: see text] or [Formula: see text] for any integer [Formula: see text] when only one GPGPU is considered, and [Formula: see text] or [Formula: see text] for [Formula: see text] GPGPUs, where [Formula: see text] is an arbitrary small value which corresponds to the target accuracy of a binary search. The proposed method is based on a dual approximation scheme that uses a dynamic programming algorithm. The associated computational costs are for the first (resp. second) family in [Formula: see text] (resp. [Formula: see text]) per step of dual approximation. The greater the value of parameter [Formula: see text], the better the approximation, but the more expensive the computational cost. Finally, we propose a relaxed version of the algorithm which achieves a running time in [Formula: see text] with a constant approximation bound of [Formula: see text]. This last result is compared to the state-of-the-art algorithm HEFT. The proposed solving method is the first general purpose algorithm for scheduling on hybrid machines with a theoretical performance guarantee that can be used for practical purposes.

Heuristic approximation and computational algorithms for closed networks: A case study in open-pit mining

We investigate a fundamental model from open-pit mining which is a cyclic system consisting of an (unreliable) shovel, trucks travelling loaded, unloading facility, and trucks travelling back empty. The interaction of these subsystems determines the mean number of trucks loaded per time unit — the capacity of the shovel, which is a fundamental quantity of interest. To determine this capacity we need the stationary probability that the shovel is idle. Because an exact analysis of the performance of the system is out of reach, besides of simulations there are various approximation algorithms proposed in the literature, which stem from computer science and can be characterized as general purpose algorithms.We propose for solving the special problem under mining conditions an extremely simple alternative algorithm. Comparison with several general purpose algorithms shows that for realistic situations in the open-pit mining application the special algorithm outperforms the precision of general purpose algorithms. This holds even if the general purpose algorithms incorporate more details of the underlying models than our simple algorithm, which is based on a strongly reduced model. The comparison and assessment is done with extensive simulations on a level of detail which the general purpose algorithms are able to cover.We discuss the application of our proposed algorithms to other applications. It turns out that our algorithms are analogues to Norton’s Theorem for a large class of general transportation systems.

Неинвазивное оценивание уровня артериального давления с помощью кардиомонитора CardioQvark

The outcome of the research on possibility to non-invasively estimate systolic blood pressure is presented. The estimating was performed by applying machine learning techniques to the data acquired with the cardiac monitor CardioQvark. The developed in Russia cardiac monitor represents a portable device capable of registering synchronous electrocardiogram and photoplethysmogram. The presented results confirm the possibility of constructing algorithms capable of estimating systolic blood pressure of individual patients. Also the possibility to construct general purpose algorithms, i.e. algorithms capable of estimating blood pressure of any patient without additional setup, was confirmed.

Combining Fragment-Ion and Neutral-Loss Matching during Mass Spectral Library Searching: A New General Purpose Algorithm Applicable to Illicit Drug Identification.

A mass spectral library search algorithm that identifies compounds that differ from library compounds by a single "inert" structural component is described. This algorithm, the Hybrid Similarity Search, generates a similarity score based on matching both fragment ions and neutral losses. It employs the parameter DeltaMass, defined as the mass difference between query and library compounds, to shift neutral loss peaks in the library spectrum to match corresponding neutral loss peaks in the query spectrum. When the spectra being compared differ by a single structural feature, these matching neutral loss peaks should contain that structural feature. This method extends the scope of the library to include spectra of "nearest-neighbor" compounds that differ from library compounds by a single chemical moiety. Additionally, determination of the structural origin of the shifted peaks can aid in the determination of the chemical structure and fragmentation mechanism of the query compound. A variety of examples are presented, including the identification of designer drugs and chemical derivatives not present in the library.

Robust Adaptive Routing Under Uncertainty

We consider the problem of finding an optimal history-dependent routing strategy on a directed graph weighted by stochastic arc costs when the objective is to minimize the risk of spending more than a prescribed budget. To help mitigate the impact of the lack of information on the arc cost probability distributions, we introduce a robust counterpart where the distributions are only known through confidence intervals on some statistics such as the mean, the mean absolute deviation, and any quantile. Leveraging recent results in distributionally robust optimization, we develop a general-purpose algorithm to compute an approximate optimal strategy. To illustrate the benefits of the robust approach, we run numerical experiments with field data from the Singapore road network. The e-companion is available at https://doi.org/10.1287/opre.2017.1662 .

Multi-threaded compression of Earth observation time-series data

ABSTRACTEarth observation data are typically compressed using general-purpose single-threaded compression algorithms that operate at a fraction of the bandwidth of modern storage and processing systems. We present evidence that recently developed multi-threaded compression codecs offer substantial benefits over widely used single-threaded codecs in terms of compression efficiency when applied to a selection of moderate resolution imaging spectroradiometer (MODIS) datasets stored in the HDF5 format. Compression codecs from the LZ77 and Rice families are shown to vary in efficacy when applied to different MODIS data products, highlighting the need for compression strategies to be tailored to different classes of data. We also introduce LPC-Rice, a new multi-threaded codec, that performs particularly well when applied to time-series data.

Contact Friction Simulating between Two Rock Bodies Using ANSYS

The ANSYS® Finite Element Method (FEM) program offers a variety of elements designed to treat cases of changing mechanical contact between the parts of an assembly or between different faces of a single part. These elements range from simple, limited idealizations to complex and sophisticated, general purpose algorithms. Contact problems are highly nonlinear and require significant computer resources to solve. Recently, analysts and designers have begun to use numerical simulation alone as an acceptable mean of validation employing numerical Finite Element Method (FEM). Contact problems fall into two general classes: rigid-to-flexible and flexible-to-flexible. In general, any time a soft material comes in contact with a hard material, the problem may be assumed to be rigid-to-flexible. The other class, flexible-to-flexible, is the more common type. To model a contact problem, you first need to identify the parts to be analyzed for their possible interaction. If one of the interactions is at a point, the corresponding component of your model is a node. If one of the interactions is at a surface, the corresponding component of your model is an element. The finite element model recognizes possible contact pairs by the presence of specific contact elements. These contact elements are overlaid on the parts of the model that are being analyzed for interaction. This paper present a simulation contact friction between Two Rock bodies loaded under two types of load condition: Axial pressure Load “σ” and Tangential Load “τ”. ANSYS® software has been used to perform the numerical calculation in this paper.

Computational aspects of assigning agents to a line

We consider the problem of assigning agents to slots on a line, where only one agent can be served at a slot and each agent prefers to be served as close as possible to his target. We introduce a general approach to compute aggregate gap-minimizing assignments, as well as gap-egalitarian assignments. The approach relies on an algorithm which is shown to be faster than general purpose algorithms for the assignment problem. We also extend the approach to probabilistic assignments and explore the computational features of existing, as well as new, methods for this setting.

Efficient Parameter Importance Analysis via Ablation with Surrogates

To achieve peak performance, it is often necessary to adjust the parameters of a given algorithm to the class of problem instances to be solved; this is known to be the case for popular solvers for a broad range of AI problems, including AI planning, propositional satisfiability (SAT) and answer set programming (ASP). To avoid tedious and often highly sub-optimal manual tuning of such parameters by means of ad-hoc methods, general-purpose algorithm configuration procedures can be used to automatically find performance-optimizing parameter settings. While impressive performance gains are often achieved in this manner, additional, potentially costly parameter importance analysis is required to gain insights into what parameter changes are most responsible for those improvements. Here, we show how the running time cost of ablation analysis, a well-known general-purpose approach for assessing parameter importance, can be reduced substantially by using regression models of algorithm performance constructed from data collected during the configuration process. In our experiments, we demonstrate speed-up factors between 33 and 14 727 for ablation analysis on various configuration scenarios from AI planning, SAT, ASP and mixed integer programming (MIP).

Accurate Small Tail Probabilities of Sums of iid Lattice-Valued Random Variables via FFT

ABSTRACTAccurately computing very small tail probabilities of a sum of independent and identically distributed lattice-valued random variables is numerically challenging. The only general purpose algorithms that can guarantee the desired accuracy have a quadratic runtime complexity that is often too slow. While fast Fourier transform (FFT)-based convolutions have an essentially linear runtime complexity, they can introduce overwhelming roundoff errors. We present sisFFT (segmented iterated shifted FFT), which harnesses the speed of FFT while retaining control of the relative error of the computed tail probability. We rigorously prove the method’s accuracy and we empirically demonstrate its significant speed advantage over existing accurate methods. Finally, we show that sisFFT sacrifices very little, if any, speed when FFT-based convolution is sufficiently accurate to begin with. Supplementary material is available online.