Data-Driven Preference-Based Deep Statistical Ranking for Comparing Multi-objective Optimization Algorithms

In this paper, the application of multi-objective evolutionary algorithms to the evaluation of performance trade-offs of planar inductors, an almost ubiquitous device in radio-frequency microelectronics, is studied. The absence of appropriate stopping criteria in most evolutionary algorithms reveals to be critical in this application. A new stopping criterion based on monitoring a set of performance metrics that account for convergence and diversity is proposed and demonstrated with practical radio-frequency circuit design problems.

The Large Synoptic Survey Telescope (LSST) is under construction in Chile. To make the delivered system meet the science goals, the project defines a set of performance metrics, and constantly monitors the system performance by evaluating the metrics against their requirements. In this paper, we describe the latest updates to the comprehensive tool set we have developed for evaluating the LSST system performance, which we collectively refer to as the LSST integrated model, and recent work on utilizing these tools for system verification. We also broaden our set of performance metrics and introduce an integrated-etendue-based metrics framework, which is useful for not just system verification, but also mitigation and optimization. Most of the major metrics currently being monitored fit under this framework, including image quality, system throughput, the single-visit point source 5σdetection limit, etc. We also mointor the Point Spread Function (PSF) ellipticity, which isn't part of this metrics framework, but is an output of the integrated model.

Improving building operation by tracking performance metrics throughout the building lifecycle (BLC)

The Guided Genetic Algorithm (GCA) is a hybrid of Genetic Algorithm and Guided Local Search, a meta-heuristic search algorithm. As the search progresses, GGA modifies both the fitness function and fitness template of candidate solutions based on feedback from constraints. The fitness template is then used to bias crossover and mutation. The Radio Link Frequency Assignment Problem (RLFAP) is a class of problem that has practical relevance to both military and civil applications. In this paper, we show how GGA can be applied to the RLFAP. We focus on an abstraction of a real life military application that involves the assigning of frequencies to radio links. GGA was tested on a set of eleven benchmark problems provided by the French military. This set of problems has been studied intensively by a number of prominent groups in Europe. It covers a variety of needs in military applications, including the satisfaction of constraints, finding optimal solutions that satisfy all the constraints and optimization of some objective functions whenever no solution exist (“partial constraint satisfaction”). Not only do these benchmark problems vary in problem nature, they are reasonably large for military applications (up to 916 variables, and up to 5548 constraints). This makes them a serious challenge to the generality, reliability as well as efficiency of algorithms. We show in this paper that GGA is capable of producing excellent results reliably in the whole set of benchmark problems.

Multi-objective evolutionary algorithms based on the summation of normalized objectives and diversified selection

Benchmark problems for numerical implementations of phase field models

Understanding the problem space in single-objective numerical optimization using exploratory landscape analysis

The existence of multi-dimensional tradeoffs among performance metrics is an impediment to the development of reliable Electronic Design Automation (EDA) tools for the design of analog circuits. To overcome this limitation, soft computing techniques that have functionality similar to the human brain are being explored. This work proposes a methodology for the development of a fast and accurate Artificial Neural Network (ANN) model for analog circuits, to achieve optimization of its design, considering a 2-stage op-amp for illustration. In order to map complex and multiple input-output relationships of analog circuits, an ANN architecture of a single hidden layer with its neuron count selected to be between the number of neurons in the input and output layers is proposed. The ANN model is developed in the MATLAB environment, and trained using SPICE-generated performance dataset. For a given set of performance metrics, the ANN model generated designs are validated against SPICE simulations. It is demonstrated that trained ANN models substitute the circuit simulator, to achieve design optimization, in a single iteration, for a new set of design specifications.

To describe performance of a mobile robot requires non-trivial efforts since robot navigation itself comprises various kinds of sub-components and the performance thereof is highly influenced by the surrounding environment and a given task. In this paper, we focus on performance metrics for robot navigation which are intended for indoor service robots. More particularly, we propose a set of performance metrics that may be useful to describe various attributes of robot navigation intended for indoor educational robots. Localization, mapping, and path planning function are considered.

Various real-world problems are essentially multiobjective optimization problems (MOPs), which involve several conflicting objectives. We propose a Quantum-inspired multiobjective Salp Swarm Algorithm based on the Decomposition technique to locate the multiple Pareto-optimal solutions (POS). The main objective while designing the algorithms for the multiobjective optimization problems is to attain a good convergence and uniform dissemination of the solutions, which remains a significant challenge for the algorithms. The proposed Decomposition-based Quantum-inspired Salp Swarm Algorithm for Multiobjective Optimization (DMQSSA) extends the primary form of SSA by using the quantum-inspired framework and a basic decomposition strategy to improve the balance between exploration and exploitation for MOPs. The Delta potential-well model (DPWM) from quantum mechanics is known for enhancing the convergence and diversity in the population, and the decomposition strategy is proved to be effective to generate evenly distributed solutions set on the Pareto front for simultaneous optimization of the subproblems. In this paper, the existing DPWM model is analysed and redesigned for MOPs with a modification in the contraction equation, and decomposition strategy is used along with an intelligent selection technique to ensure non-dominated solutions. The proposed hybrid approach is evaluated and compared with other techniques on a set of well-known benchmark problems. The results show that DMQSSA can handle the multiobjective optimization problems to find better and well-distributed Pareto optimal set. Also, success of the proposed algorithm is further illustrated on a real-world application.

In this paper, a hybrid algorithm by combining 15 task assignment rules and ant colony optimization (ACO) algorithm to solve straight and U-type assembly line balancing problem to optimize line efficiency and smoothness index is proposed. The algorithm is designed to solve assembly line balancing problems of all sizes. The proposed multi-rule multiobjective ant colony optimization algorithm for straight and U-type assembly line balancing problems is evaluated with various set of benchmark problems and compared with a multi-objective simulated annealing algorithm reported in the literature. The results indicate the better performance of the proposed hybrid algorithm.

When designing a benchmark problem set, it is important to create a set of benchmark problems that are a good generalization of the set of all possible problems. One possible way of easing this difficult task is by using artificially generated problems. In this paper, one such single-objective continuous problem generation approach is analyzed and compared with the COCO benchmark problem set, a well know problem set for benchmarking numerical optimization algorithms. Using Exploratory Landscape Analysis and Singular Value Decomposition, we show that such representations allow us to further explore the relations between the problems by applying visualization and correlation analysis techniques, with the goal of decreasing the bias in benchmark problem assessment.

A reconsideration on the resolution of phase stability analysis using stochastic global optimization methods: Proposal of a reliable set of benchmark problems

Several stakeholders are increasingly interested in operational risks and their impact on key business projects such as outsourcing projects. To guide stakeholders and project managers to take decisions related to the outsourcing of one or multiple activities, we propose a decision support system for classifying organizational risks, as example of operational risks, into different risk categories. The classification system is based on a set of performance metrics that measure both the likelihood and impact criterias. Electre-Tri a formal multi-criteria decision analysis method, was used as the foundation for this task. This method allowed us to cluster various organizational risks in different risk categories based on our current knowledge of those risks, and based on the best professional judgments. A numerical application based on a real case is presented.

Multiobjective multi-verse optimization algorithm to solve combined economic, heat and power emission dispatch problems