Proceedings of the Genetic and Evolutionary Computation Conference. Genetic and Evolutionary Computation Conference最新文献_第4页

How Fitness Aggregation Methods Affect the Performance of Competitive CoEAs on Bilinear Problems 适应度聚合方法如何影响竞争coea在双线性问题上的性能

Proceedings of the Genetic and Evolutionary Computation Conference. Genetic and Evolutionary Computation Conference Pub Date : 2023-07-12 DOI: 10.1145/3583131.3590506

Mario Alejandro Hevia Fajardo, P. Lehre

{"title":"How Fitness Aggregation Methods Affect the Performance of Competitive CoEAs on Bilinear Problems","authors":"Mario Alejandro Hevia Fajardo, P. Lehre","doi":"10.1145/3583131.3590506","DOIUrl":"https://doi.org/10.1145/3583131.3590506","url":null,"abstract":"Competitive co-evolutionary algorithms (CoEAs) do not rely solely on an external function to assign fitness values to sampled solutions. Instead, they use the aggregation of outcomes from interactions between competing solutions allowing to rank solutions and make selection decisions. This makes CoEAs a useful tool for optimisation problems that have intrinsically interactive domains. Over the past decades, many ways to aggregate the outcomes of interactions have been considered. At the moment, it is unclear which of these is the best choice. Previous research is fragmented and most of the fitness aggregation methods (fitness measures) proposed have only been studied empirically. We argue that a proper understanding of the dynamics of CoEAs and their fitness measures can only be achieved through rigorous analysis of their behaviour. In this work we make a step towards this goal by using runtime analysis to study two commonly used fitness measures. We show a dichotomy in the behaviour of a (1, Λ) CoEA when optimising a Bilinear problem. The algorithm finds a Nash equilibrium efficiently if the worst interaction is used as a fitness measure but it takes exponential time w.o.p. if the average of all interactions is used instead.","PeriodicalId":89841,"journal":{"name":"Proceedings of the Genetic and Evolutionary Computation Conference. Genetic and Evolutionary Computation Conference","volume":"37 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"80761076","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 1

Multi-objective Robust Optimization and Decision-Making Using Evolutionary Algorithms 基于进化算法的多目标鲁棒优化与决策

Proceedings of the Genetic and Evolutionary Computation Conference. Genetic and Evolutionary Computation Conference Pub Date : 2023-07-12 DOI: 10.1145/3583131.3590420

Deepanshu Yadav, Palaniappan Ramu, K. Deb

{"title":"Multi-objective Robust Optimization and Decision-Making Using Evolutionary Algorithms","authors":"Deepanshu Yadav, Palaniappan Ramu, K. Deb","doi":"10.1145/3583131.3590420","DOIUrl":"https://doi.org/10.1145/3583131.3590420","url":null,"abstract":"Evolutionary multi-objective optimization (EMO) algorithms are predominantly used for solving multi- and many-objective optimization problems to arrive at the respective Pareto front. From a practical point of view, it is desirable for a decision-maker (DM) to consider objective vectors that are less sensitive to the small perturbation in design variables and problem parameters. Such insensitive, yet closer to Pareto-optimal solutions, lie on the so-called robust front. In real-world applications, such as engineering design and process optimization problems, perturbations in variables come from manufacturing tolerances, uncertainties in material properties, variations in operating conditions, etc. The existing EMO literature on robustness studies emphasized on finding the entire robust front, but hardly considered robustness in both optimization and decision-making tasks. In this paper, we propose and evaluate different algorithmic implementations of three aspects - multi-objective optimization, robustness consideration, and multi-criterion decision-making - together. Based on experimental results on two to eight-objective problems, we discuss the outcomes and advantages of different integration approaches of these three aspects and present the most effective combined approach. The results are interesting and should pave the way to develop more efficient multi-objective robust optimization and decision-making (MORODM) procedures for handling practical problems with uncertainties.","PeriodicalId":89841,"journal":{"name":"Proceedings of the Genetic and Evolutionary Computation Conference. Genetic and Evolutionary Computation Conference","volume":"28 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87303429","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 1

AI for Scientific Discovery and a Sustainable Future 人工智能促进科学发现和可持续未来

Proceedings of the Genetic and Evolutionary Computation Conference. Genetic and Evolutionary Computation Conference Pub Date : 2023-07-12 DOI: 10.1145/3583131.3603396

C. Gomes

引用次数: 0

MOAZ: A Multi-Objective AutoML-Zero Framework 一个多目标自动零框架

Proceedings of the Genetic and Evolutionary Computation Conference. Genetic and Evolutionary Computation Conference Pub Date : 2023-07-12 DOI: 10.1145/3583131.3590391

Ritam Guha, Wei Ao, Stephen Kelly, Vishnu Boddeti, Erik D. Goodman, Wolfgang Banzhaf, Kalyanmoy Deb

{"title":"MOAZ: A Multi-Objective AutoML-Zero Framework","authors":"Ritam Guha, Wei Ao, Stephen Kelly, Vishnu Boddeti, Erik D. Goodman, Wolfgang Banzhaf, Kalyanmoy Deb","doi":"10.1145/3583131.3590391","DOIUrl":"https://doi.org/10.1145/3583131.3590391","url":null,"abstract":"Automated machine learning (AutoML) greatly eases human efforts in architecture engineering. However, mainstream AutoML methods like neural architecture search (NAS) are customized for well-designed search spaces wherein promising architectures are densely distributed. In contrast, AutoML-Zero builds machine-learning algorithms using basic primitives and can explore novel architectures beyond human knowledge. AutoML-Zero shows the potential to deploy machine learning systems by not taking advantage of either feature engineering or architectural engineering. In its current form, it only optimizes a single objective like accuracy and has no mechanism to ensure that the constraints of real-world applications are satisfied. We propose a multi-objective variant of AutoML-Zero called MOAZ, that distributes solutions on a Pareto front by trading off accuracy against the computational complexity of the machine learning algorithm. In addition to generating different Pareto-optimal solutions, MOAZ can effectively explore the sparse search space to improve search efficiency. Experimental results on linear regression tasks show MOAZ reduces the median complexity by 87.4% compared to AutoML-Zero while accelerating the median target performance achievement speed by 82%. In addition, our preliminary results on non-linear regression tasks show the potential for further improvements in search accuracy and for reducing the need for human intervention in AutoML.","PeriodicalId":89841,"journal":{"name":"Proceedings of the Genetic and Evolutionary Computation Conference. Genetic and Evolutionary Computation Conference","volume":"53 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"91328134","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Aggregation Through Adaptive Random Walks in a Minimalist Robot Swarm 基于自适应随机行走的极简机器人群聚集

Proceedings of the Genetic and Evolutionary Computation Conference. Genetic and Evolutionary Computation Conference Pub Date : 2023-07-12 DOI: 10.1145/3583131.3590485

Luigi Feola, Antoine Sion, V. Trianni, A. Reina, E. Tuci

引用次数: 0

Directed Quick Search Guided Evolutionary Algorithm for Large-scale Multi-objective Optimization Problems 大规模多目标优化问题的定向快速搜索导向进化算法

Proceedings of the Genetic and Evolutionary Computation Conference. Genetic and Evolutionary Computation Conference Pub Date : 2023-07-12 DOI: 10.1145/3583131.3590480

Ying Wu, Na Yang, Long Chen, Ye Tian, Zhenzhou Tang

{"title":"Directed Quick Search Guided Evolutionary Algorithm for Large-scale Multi-objective Optimization Problems","authors":"Ying Wu, Na Yang, Long Chen, Ye Tian, Zhenzhou Tang","doi":"10.1145/3583131.3590480","DOIUrl":"https://doi.org/10.1145/3583131.3590480","url":null,"abstract":"For large-scale multi-objective evolutionary algorithms (LSMOEAs), it has been a major challenge to efficiently obtain accurate evolutionary directions in the ultra-high-dimensional decision space to produce high-quality offspring. To this hand, this paper proposes an algorithm, namely, a directed quick search guided large-scale multi-objective evolutionary algorithm (QSLMOA). This algorithm contains three main parts: a directional vector-based sampling strategy, a quick search guided directed reproduction strategy, and an environment selection. In each generation, the proposed sampling strategy determines a set of directional solutions to construct the direction vectors which are used to guide the search directions. The sampling strategy significantly reduces the search space and improves the sampling efficiency in the early stages. On the other hand, our proposed reproduction strategy introduces the directional information, and with their assistance, the inferior ones of solutions in the combined population can rapidly converge to the elite ones, which can speed up the convergence rate. Finally, the elitist non-dominated sorting is adopted as the environment selection to obtain the parent population of the next generation. Comprehensive experiments verify that QSLMOA performs the best compared to the state-of-the-art LSMOEAs for nine large-scale multi-objective benchmark problems LSMOP1-LSMOP9 with up to three objectives and 5000 decision variables.","PeriodicalId":89841,"journal":{"name":"Proceedings of the Genetic and Evolutionary Computation Conference. Genetic and Evolutionary Computation Conference","volume":"209 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"72818639","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 1

OmnImage: Evolving 1k Image Cliques for Few-Shot Learning OmnImage:进化1k图像团为少数镜头学习

Proceedings of the Genetic and Evolutionary Computation Conference. Genetic and Evolutionary Computation Conference Pub Date : 2023-07-12 DOI: 10.1145/3583131.3590430

Lapo Frati, Neil Traft, Nick Cheney

{"title":"OmnImage: Evolving 1k Image Cliques for Few-Shot Learning","authors":"Lapo Frati, Neil Traft, Nick Cheney","doi":"10.1145/3583131.3590430","DOIUrl":"https://doi.org/10.1145/3583131.3590430","url":null,"abstract":"Few-shot learning datasets contain a large number of classes with only a few examples in each. Existing datasets may contain thousands of classes, but very simple images (e.g. handwritten characters) such that a naive baseline can perform very well. Or they may be so complex, with large within-class variation and distracting background, that they are too difficult to enable meaningful learning with so few examples. To construct a customizable dataset of consistent natural images, we assemble a new dataset with each class containing a small subset from each of the 1000 classes of ImageNet-1k. To select subsets with clear within-class consistency we use an evolutionary approach to minimize the pairwise cosine-distance between features generated by a pre-trained VGG model. We train a classifier on these evolved image cliques and find that our evolved dataset provides a greater challenge than hand written digits, but not the extreme difficulty of a non-evolved subset of ImageNet. We find that pre-training our classifier on these evolved prototypical classes significantly improves performance on classifying random subsets ImageNet (relative to pre-training on similar random-class subsets), and conjecture that these prototypical classes may be beneficial for seeding concept learning. Dataset and code is publicly available at: https://github.com/lfrati/OmnImage","PeriodicalId":89841,"journal":{"name":"Proceedings of the Genetic and Evolutionary Computation Conference. Genetic and Evolutionary Computation Conference","volume":"53 3 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79717449","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 1

To slide or not to slide? Moving along fitness levels and preserving the gene subsets diversity in modern evolutionary computation 滑还是不滑?在现代进化计算中沿着适应度水平移动并保持基因子集的多样性

Proceedings of the Genetic and Evolutionary Computation Conference. Genetic and Evolutionary Computation Conference Pub Date : 2023-07-12 DOI: 10.1145/3583131.3590491

M. Przewozniczek, M. Komarnicki

引用次数: 0

Novelty Seeking Multiagent Evolutionary Reinforcement Learning 寻求新奇的多智能体进化强化学习

Proceedings of the Genetic and Evolutionary Computation Conference. Genetic and Evolutionary Computation Conference Pub Date : 2023-07-12 DOI: 10.1145/3583131.3590428

Ayhan Alp Aydeniz, R. Loftin, Kagan Tumer

引用次数: 0

Local Optima Markov Chain: A New Tool for Landscape-aware Analysis of Algorithm Dynamics 局部最优马尔可夫链:景观感知算法动力学分析的新工具

Proceedings of the Genetic and Evolutionary Computation Conference. Genetic and Evolutionary Computation Conference Pub Date : 2023-07-12 DOI: 10.1145/3583131.3590422

F. Chicano, G. Ochoa, B. Derbel, Lorenzo Canonne

{"title":"Local Optima Markov Chain: A New Tool for Landscape-aware Analysis of Algorithm Dynamics","authors":"F. Chicano, G. Ochoa, B. Derbel, Lorenzo Canonne","doi":"10.1145/3583131.3590422","DOIUrl":"https://doi.org/10.1145/3583131.3590422","url":null,"abstract":"Landscape analysis is a very useful tool in optimization to understand the structure of the search space of a problem when there is some kind of distance or neighborhood defined over the solutions. Local Optima Networks (LON) have been proposed to serve as a summary of the landscape of a problem. LONs are graphs where the nodes are the local optima of the search space according to a particular neighborhood and edges join local optima when one can be reached from the other using some kind of perturbation followed by hill climbing. In this paper we enhance local optima networks to include precise information on the transition probabilities among local optima, yielding a Markov Chain for the visited local optima during the search. The new analysis tool, called Local Optima Markov Chain (LOMA), is built on top of the static landscape information depending on the problem and includes information about algorithm dynamics. We show how LOMAs can be used to compute metrics that are out of the reach of other landscape-aware tools, thus offering more information to understand algorithm dynamics.","PeriodicalId":89841,"journal":{"name":"Proceedings of the Genetic and Evolutionary Computation Conference. Genetic and Evolutionary Computation Conference","volume":"1 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76347621","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0