Optimization framework for battery-powered sensor selection for sound source detection

Adaptive genetic algorithms applied to dynamic multiobjective problems

This paper addresses the classic online Steiner tree problem on edge-weighted graphs. It is known that a greedy (nearest neighbor) online algorithm has a tight competitive ratio for wide classes of graphs, such as trees, rings, any class including series-parallel graphs, and unweighted graphs with bounded diameter. However, we do not know any greedy or non-greedy tight deterministic algorithm for other classes of graphs. In this paper, we observe that a greedy algorithm is $$\Omega (\log n)$$ -competitive on outerplanar graphs, where n is the number of vertices, and propose a 5.828-competitive deterministic algorithm on outerplanar graphs. Our algorithm connects a requested vertex and the tree constructed thus far using a path that is constant times longer than the distance between them. We also present a lower bound of 4 for arbitrary deterministic online Steiner tree algorithms on outerplanar graphs.

The process of regression testing is generally done under critical time and resource constraints. In the existing approaches, Test Case Prioritization (TCP), Test Case Selection (TCS), and Test Suite Reduction (TSR) have been treated as different optimization problems. This is based on simple heuristics that do not use valuable information that can be extracted from the program under test. Moreover, they do not allow for making internally consistent budget-conscious decisions. This paper introduces a unified approach that combines various variants of Graph Neural Networks (GNNs) and Reinforcement Learning (RL) for the solutions of TCP, TCS, and TSR. In our proposal, the representation of regression artifacts such as test cases, code entities, and faults is used as nodes in the typed graph, while the edges are used to show the relations between these artifacts. The relation-aware GNN is used for generating test case embeddings, reflecting the distance between the test cases and the changed areas as well as the areas known to have issues in the past. The actor-critic RL agent uses these test case embeddings and the budget to decide whether to run, skip, or discard test cases. The performance of our proposal outstrips coverage-based heuristics, history-based ranking, a GA-based search, an enhanced QPSO, and two learning-based ablations, as validated by our experiments on four Java projects from Defects4J. In terms of various programs and budgets, using GNN–RL has led to an improvement in the Average Percentage of Faults Detected (APFD) and Cost-cognizant Average Percentage of Faults Detected (APFDc) by approximately 6–9 percentage points on lower budgets. In addition, it is possible to obtain a 42% suite reduction and a 48% cost reduction while still retaining 98% of all fault-revealing tests in the final suite. The results show that it is possible to obtain promising solutions for adaptive regression testing using budget-aware reinforcement learning and graph-based representation learning.

The image formation process of a high-resolution space-borne synthetic aperture radar is a demanding task with high computation and precision requirement. When it is implemented on a restricted onboard real-time DSP processing system, the algorithm and software optimizations are essential to the success of the implementation. We implemented the capability of real-time processing for the raw data of a type of high-resolution space-borne SAR on an onboard DSP processing system. The refined Chirp Scaling algorithm is parallelized and mapped into the system. Approaches have been presented to improve the computation speed and precision, based on academic and mathematic analysis. Additionally, the formula of the refined Chirp Scaling is re-deduced, and the system is tested and verified by the simulated space-borne SAR and Radarsat-I raw data.

Laser aided deposition is a material additive based manufacturing process by metallurgically bonding the deposited material to the substrate. For a precise study of the process, a mathematical model was established to investigate thermal and mass transportation phenomena, which includes substrate melting and solidification as well as the powder heating process. The detailed process together with the temperature distribution is simulated. The simulation results of the melt pool length, width and peak temperature are validated for the laser deposition process through the measurements on the melt pool geometry and surface peak temperature by the CMOS camera and the dual-wavelength temperature sensor. A smart embedded vision system using CMOS camera is used to measure the size of the melt pool. The length and width of the melt pool are measured in real time using CMOS camera with an on-board DSP processing system which processes the images in real time and provides the output through a serial port to a real-time controller. A dual wavelength non-contact temperature sensor is used for measuring the temperature of the melt pool in real time. The comparison of the melt pool length and peak temperature was conducted between simulation and experiment and shows a good agreement.Laser aided deposition is a material additive based manufacturing process by metallurgically bonding the deposited material to the substrate. For a precise study of the process, a mathematical model was established to investigate thermal and mass transportation phenomena, which includes substrate melting and solidification as well as the powder heating process. The detailed process together with the temperature distribution is simulated. The simulation results of the melt pool length, width and peak temperature are validated for the laser deposition process through the measurements on the melt pool geometry and surface peak temperature by the CMOS camera and the dual-wavelength temperature sensor. A smart embedded vision system using CMOS camera is used to measure the size of the melt pool. The length and width of the melt pool are measured in real time using CMOS camera with an on-board DSP processing system which processes the images in real time and provides the output through a serial port to a real-t...

Efficient and robust sound source recognition and localization is one of the basic techniques for humanoid robots in terms of reaction to environments. Due to the fixed sensor arrays and limited computation resources in humanoid robots, there comes challenge for sound source recognition and localization. This article proposes a sound source recognition and localization framework to realize real-time and precise sound source recognition and localization system using humanoid robots’ sensor arrays. The type of the audio is recognized according to the cross-correlation function. And steered response power-phase transform function in discrete angle space is used to search the sound source direction. The sound source recognition and localization framework presents a new multi-robots collaboration system to get the precise three-dimensional sound source position and introduces a distance weighting revision way to optimize the localization performance. Additionally, the experiment results carried out on humanoid robot NAO demonstrate that the proposed approaches can recognize and localize the sound source efficiently and robustly.

Motivated by the problem of WDM routing in all-optical networks, we study the following NP-hard problem. We are given a directed binary tree T and a set R of directed paths on T. We wish to assign colors to paths in R, in such a way that no two paths that share a directed arc of T are assigned the same color and that the total number of colors used is minimized. Our results are expressed in terms of the depth of the tree and the maximum load l of R, i.e., the maximum number of paths that go through a directed arc of T. So far, only deterministic greedy algorithms have been presented for the problem. The best known algorithm colors any set R of maximum load l using at most 5l/3 colors. Alternatively, we say that this algorithm has performance ratio 5/3. It is also known that no deterministic greedy algorithm can achieve a performance ratio better than 5/3. In this paper we define the class of greedy algorithms that use randomization. We study their limitations and prove that, with high probability, randomized greedy algorithms cannot achieve a performance ratio better than 3/2 when applied to binary trees of depth Ω(l), and 1.293 - o(1) when applied to binary trees of constant depth. Exploiting inherent properties of randomized greedy algorithms, we obtain the first randomized algorithm for the problem that uses at most 7l/5 + o(l) colors for coloring any set of paths of maximum load l on binary trees of depth o(l1/3), with high probability. We also present an existential upper bound of 7l/5 + o(l) that holds on any binary tree. In the analysis of our bounds we use tail inequalities for random variables following hypergeometrical probability distributions which may be of their own interest.

MOGBO: A new Multiobjective Gradient-Based Optimizer for real-world structural optimization problems

This study examines the implementation of Test Case Prioritization (TCP) using the Greedy Algorithm (GA) to enhance regression testing efficiency within a financial technology company's software development cycle. With testing durations increasing significantly, this study aims to address inefficiencies by applying the Greedy Algorithm to optimize test suite size and fault detection. The research methodology involves applying the Greedy Algorithm during the Regression Testing phase, comparing the prioritized suite with the original suite using metrics such as Average Percentage Fault Detection (APFD) and Test Suite Size Reduction (TSSR). Results show that the Greedy Algorithm achieved substantial improvements in both test suite size and fault detection effectiveness across different projects. For Project A, the test suite was reduced from 51-22 test cases, achieving a TSSR of 56.8%, with an APFD increase of 206.21%, rising from 0.0853-0.2613. Project B demonstrated even greater optimization, reducing the test suite from 36-8 test cases, resulting in a TSSR of 77.8% and an APFD improvement of 83.92%, rising from 0.3194-0.5875. These outcomes underscore the algorithm’s effectiveness in eliminating redundant test cases, accelerating testing, and enhancing fault detection thereby supporting the company's goal of faster release cycles without compromising quality.

Simultaneous optimization of two or more objectives is an instance of multi-objective optimization (MOO). However, for most of the Multi-Objective Problems (MOPs), no single solution can optimize all the objective functions simultaneously. Evolutionary algorithms are a type of optimization algorithm used for constructing a well-distributed optimal front more quickly than a much more efficient approach. One of the most commonly used algorithms in this regard is differential evolution (DE). To solve the problem of non-dominated sorting for Multi-Objective DE (MODE), an approach is proposed that reduces the time complexity. DE is commonly recognized as one of the best methods for solving the MOPs. Several versions of DEs have been proposed for MOPs. This article proposes a novel mutation technique, named Environmental Parameter-based Multi-Objective Differential Evolution (EP-MODE) that acquires two additional environmental parameters to preserve diversity and accelerate the convergence. The proposed approach also reduces the time complexity of non-dominated sorting for Multi-Objective DE (MODE). The proposed mutation's performance has been evaluated on DTLZ series and ZDT series benchmark test functions on the Pareto-optimal front and compared with existing multi-objective algorithms (MODE and MODE-RMO). The solution comparatively verifies that the proposed EP-MODE outperforms most of the MODEs for lower dimension functions and higher functions. HIGHLIGHTS New environmental parameters-based mutation has been introduced in a differential evolution algorithm to solve multi-objective optimization problems (EP-mode) The addition of two additional environmental parameters to EP-mode has been found to be appropriate for maintaining diversity while also accelerating the convergence rate EP-MODE evaluated on bi-objective (DTLZ series) and tri-objective (ZDT series) benchmark test functions on the Pareto-optimal front and found EP-MODE's efficiency is higher than that of the MODE algorithm and the MODE algorithm with a ranking-based mutation operator GRAPHICAL ABSTRACT

This paper proposes a modified particle swarm optimization algorithm for engineering optimization problems with constraints, in which the penalty function is employed to the traditional PSO algorithm, and at the same time adjusts the personal optimum and global optimum to make PSO being able to solve the non-linear programming problems, then the multi-objective problem can be converted into single objective problem. Moreover, the constraint term played its role in the process of generating particles, those pariticles which don't meet the constraint condition are eliminated. The actual engineering design optimization problem is tested and the results show that the multi-objective particle swarm optimization algorithm can be used to solve the multi-objective constrained optimization problem. Comparison with Genetic Algorithm confirms that the proposed algorithm can find better solutions, and converge quickly.

Although various optimization methods for solving single-objective problems have been developed in the last few decades, these methods have lost their eligibility due to the fact that today’s engineering problems are toward multiple objective optimization problems, in real applications. For single-objective optimization problems, for example, in case of a minimization problem, this value is the decision vector giving the smallest objective that can be achieved within the specified constraints. Hence the minimum decision vector within all possible (feasible) solution vectors is the so-called optimal solution and/or optimal design. However, in multi-objective optimization problems, since a different objective value is generated against each decision vector, the superiority of the solutions over each other is determined by considering the trade-off among the objective values. Therefore, the solution of multi-objective optimization problems, unlike single-objective problems, is a set of vectors rather than a single decision vector. In multi-objective optimization problems, especially if there are intricate objectives, the computational cost of the problem increases. In other words, while synchronously trying to maximize one of the objectives and to minimize another one makes it difficult to find the global optimum design. One of the important techniques used in multi-objective optimization problems is Pareto optimality which enables to select the global optimum solution taking into account the trade-off among all objectives. In this context, using of derivative-based methods has decreased, but the use of metaheuristic methods has increased due to the rapid availability of global optimum solution. This is because the improvements in the field of optimization are progressing in proportion to technology and varying according to the needs. In this chapter, one of the recent metaheuristic optimization methods based on swarm intelligence that is so-called a Pareto-based bat algorithm inspired by the behavior of determining the direction and distance of an object using the echo of the sound called the echolocation of bats is used in order to obtain optimum solutions for multi-objective engineering design problems. In this regard, a four-bar planar truss, a real-sized welded steel beam as well as a multi-layer radar absorber are selected as multi-objective engineering design optimization problems. In case the obtained results (optimal designs) are examined, the potency and the reliability of the proposed multi-objective Pareto-based bat algorithm are demonstrated.

Greedy method, as an efficient computing tool, can be applied to various combinatorial or nonlinear optimization problems where finding the global optimum is difficult, if not computationally infeasible. A greedy algorithm has the nature of making the locally optimal choice at each stage and then solving the subproblems that arise later. It iteratively makes one greedy choice after another, reducing each given problem into a smaller one. In other words, a greedy algorithm never reconsiders its choices. Clearly, greedy method often fails to find the globally optimal solution. However, a greedy algorithm can be proven to yield the global optimum for a given class of problems such as Kruskal's algorithm and Prim's algorithm for finding minimum spanning tree, Dijkstra's algorithm for finding single-source shortest path, and the algorithm for finding optimum Huffman tree [5]. Even for some optimization problems proven to be NP hard, a greedy algorithm may generate near optimal solution with high probability if one exploits the problem structure properly. In this chapter, we focus on the optimization problems arising from plume detection, localization and tracking and provide convincing argument on the usefulness of greedy algorithms. Detection, identification, localization, tracking and prediction of chemical, biological or nuclear propagation is crucial to battlefield surveillance and homeland security. In addition, post-accident management for public protection relies critically on detecting and tracing dangerous gas leakages promptly. The determination of source origins and release rates is useful for the forecast of gas concentration in the atmosphere and for the management staff to prioritize off-site evacuation plans. A lot of research has been focused on detecting and localizing single or multiple plume sources with autonomous vehicles [11] or sensor networks such as [22] for a vapor-emitting source, [2] for a nuclear source, and [14, 15] for a chemical source. In [12] the plume detection and localization problem is formulated as abrupt change detection using sparse sensor measurements. The development of a large scale testbed has been reported in [8] for plume detection, identification and tracking. In [3] dense sensor coverage has been used for radioactive source detection while [26] showed that using three error-free intensity sensors, one can identify the plume origin to any desired accuracy with high probability. Although this approach offers an effective solution with linear complexity of the hypothesis space, a major limitation is that the continuous time dynamic model of plume propagation has to be in the product form. O pe n A cc es s D at ab as e w w w .in te ch w eb .o rg

Sound source detection and localization have a lot of practical uses in many industrial settings. Most of sound source direction detection algorithms in literature are designed to identify the angle of sound source in a 2D space. In this work, we propose to use convolutional neural networks to detect the sound source direction in a 3D space. This algorithm is based on the generalized cross correlation method with phase transform (GCC-PHAT) [1] to derive time delay of arrival (TDOA). By using a convolutional neural network model, this algorithm can be applied and deployed. In addition, by modifying GCC-PHAT formula, this approach also works of multiple sound sources detection. Simulation experimental results on single sound source and multiple sound sources detection show the proposed system could work in most situations.

The voting rules proposed by Dodgson and Young are both designed to find the alternative closest to being a Condorcet winner, according to two different notions of proximity; the score of a given alternative is known to be hard to compute under either rule.In this paper, we put forward two algorithms for approximating the Dodgson score: an LP-based randomized rounding algorithm and a deterministic greedy algorithm, both of which yield an O(log m) approximation ratio, where m is the number of alternatives; we observe that this result is asymptotically optimal, and further prove that our greedy algorithm is optimal up to a factor of 2, unless problems in NP have quasi-polynomial time algorithms. Although the greedy algorithm is computationally superior, we argue that the randomized rounding algorithm has an advantage from a social choice point of view.Further, we demonstrate that computing any reasonable approximation of the ranking produced by Dodgson's rule is NP-hard. This result provides a complexity-theoretic explanation of sharp discrepancies that have been observed in the Social Choice Theory literature when comparing Dodgson elections with simpler voting rules.Finally, we show that the problem of calculating the Young score is NP-hard to approximate by any factor. This leads to an inapproximability result for the Young ranking.