This paper puts forward a proposal for combining multi-operator evolutionary algorithms (EAs), in which three EAs, each with multiple search operators, are used. During the evolution process, the algorithm gradually e...
详细信息
ISBN:
(纸本)9781479914883
This paper puts forward a proposal for combining multi-operator evolutionary algorithms (EAs), in which three EAs, each with multiple search operators, are used. During the evolution process, the algorithm gradually emphasizes on the best performing multi-operator EA, as well as the search operator. The proposed algorithm is tested on the CEC2014 single objective real-parameter competition. The results show that the proposed algorithm has the ability to reach good solutions.
Complex Valued Neural Networks play an increasing role in machine learning due to their capability to characterize intricate problems and to their appropriateness to engineering fields which use complex number or phas...
详细信息
ISBN:
(纸本)9781479968213
Complex Valued Neural Networks play an increasing role in machine learning due to their capability to characterize intricate problems and to their appropriateness to engineering fields which use complex number or phasor representations. This paper presents a new method for training the Phase-Based Neurons, a new type of Complex Valued Neural Networks that uses just the phase of the complex activation for operation and representation. The paper investigates also the capabilities this type of neural network to solve the N bit parity problem and its efficiency in doing this.
The increasing global demand for sustainable and cleaner transportation has led to extensive research on alternative fuels for Internal Combustion (IC) engines. One promising option is the utilization of methane/hydro...
详细信息
The increasing global demand for sustainable and cleaner transportation has led to extensive research on alternative fuels for Internal Combustion (IC) engines. One promising option is the utilization of methane/hydrogen blends in Spark-Ignition (SI) engines due to their potential to reduce Green House Gas (GHG) emissions and improve engine performance. However, the optimal operation of such an engine is challenging due to the interdependence of multiple conflicting objectives, including Brake Mean Effective Pressure (BMEP), Brake Specific Fuel Consumption (BSFC), and nitrogen oxide (NOx) emissions. This paper proposes an evolutionary optimization algorithm that employs a surrogate model as a fitness function to optimize methane/hydrogen SI engine performance and emissions. To create the surrogate model, we propose a novel ensemble learning algorithm that consists of several base learners. This paper employs ten different learning algorithms diversified via the Wagging method to create a pool of base-learner algorithms. This paper proposes a combinatorial evolutionary pruning algorithm to select an optimal subset of learning algorithms from a pool of base learners for the final ensemble algorithm. Once the base learners are designed, they are incorporated into an ensemble, where their outputs are aggregated using a weighted voting scheme. The weights of these base learners are optimized through a gradient descent algorithm. However, when optimizing a problem using surrogate models, the fitness function is subject to approximation uncertainty. To address this issue, this paper introduces an uncertainty reduction algorithm that performs averaging within a sphere around each solution. Experiments are performed to compare the proposed ensemble learning algorithm to the classical learning algorithms and state-of-the-art ensemble algorithms. Also, the proposed smoothing algorithm is compared with the state-ofthe-art evolutionary algorithms. Experimental studies suggest th
Synthetic aperture radar (SAR) classification models based on convolutional neural networks have high accuracy, but the models' security is still threatened by adversarial examples. The high threat of adversarial ...
详细信息
Synthetic aperture radar (SAR) classification models based on convolutional neural networks have high accuracy, but the models' security is still threatened by adversarial examples. The high threat of adversarial examples derives from the invisible noise that can cause feature changes within the model. Among many adversarial examples detection methods, feature attribution that is sensitive to feature changes performs well in feature analysis. Unfortunately, the existing feature attribution-based detection methods cannot balance the computational efficiency and detection performance well due to the size and speckle noise of the images in SAR adversarial examples detection. In this work, we propose the Dual-objective Feature Attribution (DoFA) method by using the feature attribution scan block to find the suitable scan granularity. The DoFA method formulates the SAR adversarial examples detection issue as a dual-objective optimization problem and takes the number of subsamples generated by feature analysis and the area under curve (AUC) value of the logistic regression model as the objective functions while the feature scan block's size, stride, padding, and the number of selected model layers are the decision variables. Non-dominated Sorting Genetic Algorithm-II (NSGA-II) is adopted to search for the scan block parameters with Pareto optimality so that the DoFA method can automatically obtain the best feature analysis granularity in different scenarios. The experimental results on the FUSAR-Ship dataset have shown that the proposed DoFA method has a higher AUC value under five adversarial attacks and a smaller number of subsamples than the existing adversarial examples detection method.
Genetic algorithms (GAs) are meta-heuristic algorithms that are used for solving constrained and unconstrained optimization problems, mimicking the process of natural selection in biological evolution. Due to the fact...
详细信息
Genetic algorithms (GAs) are meta-heuristic algorithms that are used for solving constrained and unconstrained optimization problems, mimicking the process of natural selection in biological evolution. Due to the fact that GAs do not require the optimization function to be differentiable, they are suitable for application in cases where the derivative of the objective function is either unavailable or impractical to obtain numerically. This paper proposes a general purpose genetic algorithm toolkit, implemented in Python3 programming language, having only minimum dependencies in NumPy and Joblib, that handle some of the numerical and parallel execution details.
Topology optimizations involving evolutionary algorithms are promising approaches to solve practical engineering design problems, since their use of derivation-free algorithms makes them applicable to any design probl...
详细信息
Topology optimizations involving evolutionary algorithms are promising approaches to solve practical engineering design problems, since their use of derivation-free algorithms makes them applicable to any design problem. In some evolutional topology optimization methods, structures are determined through the use of spatially smooth parametric functions. This article, first, presents a new parametric level set function using kernel functions as a general formulation of spatially smooth function-based methods. It is called the kernel level set function, which is used to determine the distributions of materials in the design region. By using the evolutionary algorithm to modify the profile of the kernel level set function, topology optimizations can be realized for various design problems, and the shape expression ability of the kernel level set function can be customized by changing the kernel functions. Second, a multi-objective formulation considering an objective and structural complexity is introduced, and an optimization algorithm specialized for solving it is proposed. The combination of the kernel level set function and the present optimization algorithm makes it possible to solve various topology optimization problems, and Pareto fronts with respect to target performance and structural complexity can be obtained. The present method is applied to three topology optimization problems. The numerical results are used to examine the characteristics of the present method for various kernel functions. Furthermore, the present method is applied to an optimization problem of an interior permanent magnet motor for electric vehicles. It is shown that the present method can solve practical design problems that have various objectives and constraint conditions, thanks to the use of a derivation-free evolutionary algorithm.
The operational optimization of the coal mine integrated energy system (CMIES) is crucial for reducing costs and carbon emissions. However, the system's multi-objective nature, stringent constraints, and the uncer...
详细信息
The operational optimization of the coal mine integrated energy system (CMIES) is crucial for reducing costs and carbon emissions. However, the system's multi-objective nature, stringent constraints, and the uncertainty of renewable and mine-derived energy make solving its optimization challenging. Thus, this paper first presents a data-driven uncertainty transformation method to address the uncertainty of renewable energy and mining derived energy output;then, a multi-task multi-objective evolutionary algorithm based on adaptive auxiliary tasks (MMOEA-AS) is proposed, which includes a main task and three auxiliary tasks. Meanwhile, an adaptive update strategy for auxiliary tasks and a matching degree-guided knowledge transfer mechanism are proposed to improve the performance of the algorithm. Finally, taking the energy scheduling problem of a coal mine in Shanxi, China as an example, MMOEA-AS is compared with five advanced evolutionary algorithms. The results show that MMOEA-AS can effectively solve the operation optimization of the CMIES, and obtain the optimal scheduling results.
Dynamic linear functions on the boolean hypercube are functions which assign to each bit a positive weight, but the weights change over time. Throughout optimization, these functions maintain the same global optimum, ...
详细信息
Dynamic linear functions on the boolean hypercube are functions which assign to each bit a positive weight, but the weights change over time. Throughout optimization, these functions maintain the same global optimum, and never have defecting local optima. Nevertheless, it was recently shown [Lengler, Schaller, FOCI 2019] that the (1 + 1)-evolutionary Algorithm needs exponential time to find or approximate the optimum for some algorithm configurations. In this experimental paper, we study the effect of larger population sizes for dynamic binval, the extreme form of dynamic linear functions. We find that moderately increased population sizes extend the range of efficient algorithm configurations, and that crossover boosts this positive effect substantially. Remarkably, similar to the static setting of monotone functions in [Lengler, Zou, FOGA 2019], the hardest region of optimization for (mu + 1)-EA is not close the optimum, but far away from it. In contrast, for the (mu + 1)-GA, the region around the optimum is the hardest region in all studied *** check and confirm the inserted city name is correctly ***.
Brains have evolved diverse neurons with varying morphologies and dynamics that impact temporal information processing. In contrast, most neural network models use homogeneous units that vary only in spatial parameter...
详细信息
Brains have evolved diverse neurons with varying morphologies and dynamics that impact temporal information processing. In contrast, most neural network models use homogeneous units that vary only in spatial parameters (weights and biases). To explore the importance of temporal parameters, we trained spiking neural networks on tasks with varying temporal complexity, holding different parameter subsets constant. We found that adapting conduction delays is crucial for solving all test conditions under tight resource constraints. Remarkably, these tasks can be solved using only temporal parameters (delays and time constants) with constant weights. In more complex spatio-temporal tasks, an adaptable bursting parameter was essential. Overall, allowing adaptation of both temporal and spatial parameters enhances network robustness to noise, a vital feature for biological brains and neuromorphic computing systems. Our findings suggest that rich and adaptable dynamics may be the key for solving temporally structured tasks efficiently in evolving organisms, which would help explain the diverse physiological properties of biological neurons.
Explaining the decisions made by population-based metaheuristics can often be considered difficult due to the stochastic nature of the mechanisms employed by these optimisation methods. As industries continue to adopt...
详细信息
Explaining the decisions made by population-based metaheuristics can often be considered difficult due to the stochastic nature of the mechanisms employed by these optimisation methods. As industries continue to adopt these methods in areas that increasingly require end-user input and confirmation, the need to explain the internal decisions being made has grown. In this article, we present our approach to the extraction of explanation supporting features using trajectory mining. This is achieved through the application of principal components analysis techniques to identify new methods of tracking population diversity changes post-runtime. The algorithm search trajectories were generated by solving a set of benchmark problems with a genetic algorithm and a univariate estimation of distribution algorithm and retaining all visited candidate solutions which were then projected to a lower dimensional sub-space. We also varied the selection pressure placed on high fitness solutions by altering the selection operators. Our results show that metrics derived from the projected sub-space algorithm search trajectories are capable of capturing key learning steps and how solution variable patterns that explain the fitness function may be captured in the principal component coefficients. A comparative study of variable importance rankings derived from a surrogate model built on the same dataset was also performed. The results show that both approaches are capable of identifying key features regarding variable interactions and their influence on fitness in a complimentary fashion.
暂无评论