Various evolutionary multiobjective optimization algorithms (EMOAs) have replaced or augmented the notion of dominance with quality indicators and leveraged them in selection operators. Recent studies show that indica...
详细信息
ISBN:
(纸本)9781450311779
Various evolutionary multiobjective optimization algorithms (EMOAs) have replaced or augmented the notion of dominance with quality indicators and leveraged them in selection operators. Recent studies show that indicator-based EMOAs outperform traditional dominance-based EMOAs. This paper proposes and evaluates an ensemble learning method that constructs an ensemble of existing indicators with a novel boosting algorithm called Pdi-Boosting. The proposed method is carried out with a training problem in which Pareto-optimal solutions are known. It can work with a simple training problem, and an ensemble of indicators can effectively aid parent selection and environmental selection in order to solve harder problems. Experimental results show that the proposed method is efficient thanks to its dynamic adjustment of training data. An ensemble of indicators outperforms existing individual indicators in optimality, diversity and robustness. The proposed ensemble-based evolutionary algorithm outperforms a well-known dominance-based EMOA and existing indicator-based EMOAs.
In a first-of-its-kind study, this paper formulates the problem of estimating the prediction intervals (PIs) in a macroeconomic time series as a bi-objective optimization problem and solves it with three evolutionary ...
详细信息
In a first-of-its-kind study, this paper formulates the problem of estimating the prediction intervals (PIs) in a macroeconomic time series as a bi-objective optimization problem and solves it with three evolutionaryalgorithms namely, Non-dominated Sorting Genetic Algorithm (NSGA-II), Non-dominated Sorting Particle Swarm optimization (NSPSO) and Multi-Objective evolutionary Algorithm based on Decomposition (MOEA-D). We also proposed modeling the chaos present in the time series as a preprocessor, which we called stage-1. Accordingly, we proposed 2-stage models, where stage-1 is followed by obtaining the optimal point prediction using NSGA-II/NSPSO/MOEA-D and using these point predictions to obtain PIs (stage-2). We then proposed a 3-stage hybrid, which is built on the 2-stage model, wherein the 3rd stage also invokes NSGA-II/NSPSO/MOEA-D in order to estimate the PIs from the point predictions obtained in 2nd stage by simultaneously and explicitly optimizing (i) prediction interval coverage probability (PICP) and (ii) prediction interval average width (PIAW). The proposed models yielded better results in terms of both PICP and PIAW compared to the state-of-the-art Lower Upper Bound Estimation Method (LUBE) with Gradient Descent (GD) and LUBE with long short-term memory (LSTM) network. The 3-stage models outperformed the 2-stage models with respect to PICP but showed similar performance in PIAW at the cost of running NSGA-II/NSPSO/MOEA-D second time. Overall, MOEA-D yielded best PIs in two datasets and NSGA-II outperformed the other two in the third dataset. But, in terms of hypervolume, in 2-stage MOEA-D produced most diverse solutions in two datasets, while NSGA-II was the winner in the third dataset.
Convolutional neural networks have achieved remarkable success in the field of computer vision. However, due to their high storage and expensive computations, recently, there has been a lot of work focusing on reducin...
详细信息
Convolutional neural networks have achieved remarkable success in the field of computer vision. However, due to their high storage and expensive computations, recently, there has been a lot of work focusing on reducing the complexity of convolutional neural networks. In this work, we propose a random filter pruning method by means of evolutionary multiobjective optimization algorithms to accelerate the Siamese ResNet-50 for remote sensing scene classification. We have conduct experiments on NWPU-RESISC45, UC Merced Land-Use and SIRI-WHU datasets for performance evaluation of the proposed method. The experimental results demonstrate that the classification performance of our pruned model has been improved while keeping a certain degree of sparsity of the model. (C) 2019 Elsevier B.V. All rights reserved.
Various evolutionary multiobjective optimization algorithms (EMOAs) have adopted indicator-based selection operators that augment or replace dominance ranking with quality indicators. A quality indicator measures the ...
详细信息
ISBN:
(纸本)9780769545967
Various evolutionary multiobjective optimization algorithms (EMOAs) have adopted indicator-based selection operators that augment or replace dominance ranking with quality indicators. A quality indicator measures the goodness of each solution candidate. Many quality indicators have been proposed with the intention to capture different preferences in optimization. Therefore, indicator-based selection operators tend to have biased selection pressures that evolve solution candidates toward particular regions in the objective space. An open question is whether a set of existing indicator-based selection operators can create a single operator that outperforms those existing ones. To address this question, this paper studies a method to aggregate (or boost) existing indicator-based selection operators. Experimental results show that a boosted selection operator outperforms exiting ones in optimality, diversity and convergence velocity. It also exhibits robustness against different characteristics in different optimization problems and yields stable performance to solve them.
The performance of a multiobjectiveevolutionary Algorithm (MOEA) for many-objective optimization is often evaluated by multiobjective scalable test problems like DTLZ and WFG problems. This is because the scalable te...
详细信息
ISBN:
(纸本)9781728121536
The performance of a multiobjectiveevolutionary Algorithm (MOEA) for many-objective optimization is often evaluated by multiobjective scalable test problems like DTLZ and WFG problems. This is because the scalable test problems are quite useful for an MOEA analysis. However, the scalable test problems do not have enough diversity of the shapes of the Pareto front and the feasible region to evaluate the capability of MOEAs. Previous studies showed that these shapes have a great impact on the performance of MOEAs. Thus, MOEAs should be evaluated on more test problems with different shapes of the Pareto front and the feasible region. In this study, the shapes of the Pareto front in the existing scalable test problems are examined from some viewpoints such as the distribution of optimal or worst solutions for each objective and the degree of the correspondence with the distribution of the weight vectors. The shape of the feasible region is also examined from the viewpoint of the spread of an initial population and the existence of dominance resistant solutions. According to the observations, we propose new shapes of the Pareto front and the feasible region to design a new scalable test suite. Experimental results show that the proposed test suite has totally different properties from the existing test problems.
A multiobjectiveevolutionary Algorithm (MOEA) is one of the effective approaches for solving multiobjectiveoptimization Problems (MOPs). The performance of MOEAs is evaluated mainly by scalable MOP test suites where...
详细信息
ISBN:
(纸本)9781538666500
A multiobjectiveevolutionary Algorithm (MOEA) is one of the effective approaches for solving multiobjectiveoptimization Problems (MOPs). The performance of MOEAs is evaluated mainly by scalable MOP test suites where the number of objectives can be arbitrarily specified. However, the number of scalable MOP test suites is quite limited and their properties are similar. Thus, there is a risk that the current research on MOEAs is specialized for some properties (i.e., a shape of feasible regions, a shape of the Pareto front, and a distance function) of existing scalable MOP test suites. In this paper, we focus on the above properties of two popular MOP test suites (i.e., DTLZ and WFG). Based on DTLZ and WFG, we create 12 MOPs which have partially different properties from those of DTLZ and WFG. Computational experiments show that the search performance of the state-of-the-art MOEAs strongly depends on three properties.
Molecular communication provides communication and networking capabilities for nanomachines such as biosensors and bio-actuators to form and enable Body Area NanoNetworks (BANNs). This paper considers neuron-based mol...
详细信息
Molecular communication provides communication and networking capabilities for nanomachines such as biosensors and bio-actuators to form and enable Body Area NanoNetworks (BANNs). This paper considers neuron-based molecular communication, which utilizes natural neurons as a primary component to build BANNs, and proposes an end-to-end software architecture to manage and control neuron-based BANNs through a series of software services. Those services aid to realize end user applications in healthcare, such as biomedical and rehabilitation applications. In the proposed architecture, a neuron-based BANN consists of a set of nanomachines and a network of neurons that are artificially formed into a particular topology. This paper investigates two mechanisms in the proposed architecture: (1) an artificial assembly method to form neurons into specific three-dimensional topology patterns and (2) a communication protocol for neuronal signaling based on Time Division Multiple Access (TDMA), called Neuronal TDMA. The assembly method uses silica beads as growth surface and bead-bead contacts as geometrical constraints on neuronal connectivity. A web lab experiment verifies this method with neuronal hippocampal cells. Neuronal TDMA leverages an evolutionarymultiobjectiveoptimization algorithm (EMOA) to optimize the signaling schedules for nanomachines. Simulation results demonstrate that the Neuronal TDMA efficiently obtains quality solutions.
This paper proposes and evaluates Neuronal TDMA, a TDMA-based signaling protocol framework for molecular communication, which utilizes neurons as a primary component to build in-body sensor-actuator networks (IBSANs)....
详细信息
ISBN:
(纸本)9781936968602
This paper proposes and evaluates Neuronal TDMA, a TDMA-based signaling protocol framework for molecular communication, which utilizes neurons as a primary component to build in-body sensor-actuator networks (IBSANs). Neuronal TDMA leverages an evolutionarymultiobjectiveoptimization algorithm (EMOA) that optimizes the signaling schedule for nanomachines in IBSANs. The proposed EMOA uses a population of solution candidates, each of which represents a particular signaling schedule, and evolves them via several operators such as selection, crossover, mutation and offspring size adjustment. The evolution process is performed to seek Pareto-optimal signaling schedules subject to given constraints. Simulation results verify that the proposed EMOA efficiently obtains quality solutions. It outperforms several conventional EMOAs.
Recently, the research on quantum-inspired evolutionaryalgorithms (QEA) has attracted some attention in the area of evolutionary computation. QEA use a probabilistic representation, called Q-bit, to encode individual...
详细信息
ISBN:
(纸本)9781424481262
Recently, the research on quantum-inspired evolutionaryalgorithms (QEA) has attracted some attention in the area of evolutionary computation. QEA use a probabilistic representation, called Q-bit, to encode individuals in population. Unlike standard evolutionaryalgorithms, each Q-bit individual is a probability model, which can represent multiple solutions. Since probability models store global statistical information of good solutions found previously in the search, QEA have good potential to deal with hard optimization problems with many local optimal solutions. So far, not much work has been done on evolutionary multi-objective (EMO) algorithms with probabilistic representation. In this paper, we investigate the performance of two state-of-the-art EMO algorithms MOEA/D and NSGA-II, with probabilistic representation based on pheromone trails, on the multi-objective travelling salesman problem. Our experimental results show that MOEA/D and NSGA-II with probabilistic presentation are very promising in sampling high-quality offspring solutions and in diversifying the search along the Pareto fronts.
暂无评论