检索结果-内蒙古大学图书馆

Genetic and evolutionary computation Conference (GECCO)

作者： Schneider, Lennart Bischl, Bernd Thomas, Janek Ludwig Maximilians Univ Munchen Munich Germany Munich Ctr Machine Learning MCML Munich Germany

ISBN: (纸本)9798400701191

We present a model-agnostic framework for jointly optimizing the predictive performance and interpretability of supervised machine learning models for tabular data. Interpretability is quantified via three measures: feature sparsity, interaction sparsity of features, and sparsity of non-monotone feature effects. By treating hyperparameter optimization of a machine learning algorithm as a multi-objective optimization problem, our framework allows for generating diverse models that trade off high performance and ease of interpretability in a single optimization run. Efficient optimization is achieved via augmentation of the search space of the learning algorithm by incorporating feature selection, interaction and monotonicity constraints into the hyperparameter search space. We demonstrate that the optimization problem effectively translates to finding the Pareto optimal set of groups of selected features that are allowed to interact in a model, along with finding their optimal monotonicity constraints and optimal hyperparameters of the learning algorithm itself. We then introduce a novel evolutionary algorithm that can operate efficiently on this augmented search space. In benchmark experiments, we show that our framework is capable of finding diverse models that are highly competitive or outperform state-of-the-art XGBoost or Explainable Boosting Machine models, both with respect to performance and interpretability.

关键词： supervised learning performance interpretability tabular data multi-objective evolutionary computation group structure

来源：评论

学校读者我要写书评

暂无评论

Pretraining Reduces Runtime in Denoising Autoencoder Genetic Programming by an Order of Magnitude

Pretraining Reduces Runtime in Denoising Autoencoder Genetic...

引用

Genetic and evolutionary computation Conference (GECCO)

作者： Reiter, Johannes Schweim, Dirk Wittenberg, David Univ Appl Sci Mainz Mainz Germany Johannes Gutenberg Univ Mainz Mainz Germany

ISBN: (纸本)9798400701207

Denoising autoencoder genetic programming (DAE-GP) is an estimation of distribution genetic programming (EDA-GP) algorithm. It uses denoising autoencoder long short-term memory networks as probabilistic model to replace the standard mutation and recombination operators of genetic programming (GP). Recent work has shown several advantages regarding solution length and overall performance of DAE-GP when compared to GP. However, training a neural network at each generation is computationally expensive, where model training is the most time consuming process of DAE-GP. In this work, we propose pretraining to reduce the runtime of the DAE-GP. In pretraining, the neural network is trained preceding the evolutionary search. In experiments on 8 real-world symbolic regression tasks we find that DAE-GP with pretraining has a reduced overall runtime of an order of magnitude while generating individuals with similar or better fitness.

关键词： estimation of distribution genetic programming evolutionary computation pretraining

来源：评论

学校读者我要写书评

暂无评论

Development of Deep Generalized Radial Basis Function Network for Short-term Load Forecasting 22

Development of Deep Generalized Radial Basis Function Networ...

引用

22nd World Congress of the International Federation of Automatic Control (IFAC)

作者： Miwa, Rikuto Mori, Hiroyuki Meiji Univ Dept Network Design Tokyo 1648525 Japan

ISBN: (纸本)9781713872344

In this paper, an efficient method is proposed for short-term load forecasting (STLF) in power systems. The proposed method integrates Generalized Radial Basis Function Network (GRBFN) of Artificial Neural Network (ANN) with Autoencoder of pretraining in Deep Neural Network (DNN). GRBFN is an extension of Radial Basis Function Network (RBFN) in a way that the parameters of the Gaussian function are determined by the learning process. Autoencoder plays an important role to reduce the number of input variables, which means the dimensionality reduction due to the feature extraction. The hybrid model of GRBFN and Autoencoder results in DNN to improve forecasting model accuracy. Also, evolutionary Particle Swarm Optimization (EPSO) of evolutionary computation is employed to optimize the parameters of GRBFN. The proposed method is successfully applied to real data of short-term load forecasting. Copyright (c) 2023 The Authors.

关键词： Short-term load forecasting Artificial Intelligence Artificial neural network Deep neural networks evolutionary computation evolutionary Particle Swarm Optimization (EPSO)

来源：评论

学校读者我要写书评

暂无评论

Multi-objective Baby Search Algorithm 14th

Multi-objective Baby Search Algorithm

引用

14th International Conference on Swarm Intelligence (ICSI)

作者： Liu, Yi Li, Gengsong Qin, Wei Li, Xiang Liu, Kun Wang, Qiang Zheng, Qibin Acad Mil Sci Beijing 100091 Peoples R China Natl Innovat Inst Def Technol Beijing 100071 Peoples R China

ISBN: (纸本)9783031366215;9783031366222

Multi-objective optimization problems are commonplace in real-world applications, and evolutionary algorithms are successful in solving them. Baby search algorithm is a novel evolutionary algorithm proposed recently, which has excellent ability on exploration and exploitation. However, it is designed to cater to single-objective optimization problems, but in this paper, we expand and modify it for multi-objective optimization. We introduce the boundary selection strategy to choose individuals from the Pareto archive for generating new solutions. To determine the best position of each individual we combine Pareto domination relation with random selection. Additionally, we propose an adapted Levy flight method to find promising solutions. Eleven standard multi-objective testing instances, five prevailing indicators and five state-of-art algorithms are applied to evaluate our algorithm. Experiments results demonstrate that our algorithm performs well on IGD, HV, Spread and GD measures.

关键词： Baby Search Algorithm Multi-objective Optimization evolutionary computation Heuristic Algorithm Artificial Intelligence

来源：评论

学校读者我要写书评

暂无评论

Evolution of Walsh Transforms with Genetic Programming

Evolution of Walsh Transforms with Genetic Programming

引用

Genetic and evolutionary computation Conference (GECCO)

作者： Rovito, Luigi De Lorenzo, Andrea Manzoni, Luca Univ Trieste Trieste Italy

ISBN: (纸本)9798400701207

The design of Boolean functions which exhibit high-quality cryptography properties is a crucial aspect when implementing secure stream ciphers. To this end, several methods have been proposed to search for secure Boolean functions, and, among those, evolutionary algorithms play a prominent role. In this paper, Genetic Programming (GP) is applied for the evolution of Boolean functions in order to maximize one essential property for strong cryptography functions, namely non-linearity. Differently from other approaches, the evolution happens in the space of Walsh Transforms, instead of using a direct representation of the Boolean functions. Specifically, we evolve coefficients of the Walsh Transform to obtain a generic Walsh spectrum, from which it is possible, through spectral inversion, to obtain a pseudo-Boolean function that, consequently, can be mapped to (one of) the nearest Boolean one. Since that function might not be unique, we propose a strategy in which balancedness, another important cryptography property, is preserved as much as possible. To show that the evolutionary search is actually effective in this task, we evolved Boolean functions from 6 to 16 variables. The results show that not only GP is effective in evolving Boolean functions with high non-linearity, but also that balanced functions are discovered.

关键词： evolutionary computation evolutionary Algorithms Genetic Programming Cybersecurity Cryptography Walsh Transform Spectral Inversion Boolean Functions Stream Ciphers Non-Linearity

来源：评论

学校读者我要写书评

暂无评论

Evolving Dilation Functions for Parameter Estimation 20

Evolving Dilation Functions for Parameter Estimation

引用

20th IEEE Conference on computational Intelligence in Bioinformatics and computational Biology (IEEE CIBCB)

作者： Papetti, Daniele M. Coelho, Vasco Univ Milano Bicocca Dept Informat Syst & Commun Milan Italy

ISBN: (纸本)9798350310177

Global optimization problems are among the most complex and widespread tasks in Computer Science. The capability of finding the global optimum is often hindered by many features-e.g., multi-modality, noisiness and non-differentiability-of the fitness landscape related to the problem. To overcome such issues, Dilation Functions (DFs) can be used to perform problem-dependent manipulations of the fitness landscape to "expand" promising regions and "compress" less promising regions. Since in many real-world scenarios the knowledge of the problem characteristics to handcraft tailored DFs is lacking, two automatic approaches to evolve ad-hoc DFs have been proposed and assessed on benchmark problems. One approach is a two-layered method that leverages an Evolution Strategies (ES) and a self-tuning variant of Particle Swarm Optimization to evolve a DF. The other approach uses Genetic Programming (GP) to evolve a set of tailored DFs for each dimension of the search space. In this work, we introduce evolutionary LBDFs (EvLBDFs), a novel approach based on ES to evolve Local Bubble Dilation Functions, a family of DFs that locally dilate hyper-spherical bounded regions of the search space. Moreover, we compare these approaches to solve the Parameter Estimation (PE) problems of two bio-chemical systems. Our results highlight that all three approaches evolved DFs that simplified the PE landscapes. The GP-based approach outperformed the other approaches on the PE problem with the higher number of kinetic parameters to infer.

关键词： Parameter Estimation Dilation Functions evolutionary computation Optimization Problems Local Bubble Dilation Functions

来源：评论

学校读者我要写书评

暂无评论

Evolving Graphs with Cartesian Genetic Programming with Lexicase Selection

Evolving Graphs with Cartesian Genetic Programming with Lexi...

引用

Genetic and evolutionary computation Conference (GECCO)

作者： Lavinas, Yuri Cotacero, Kevin Cussat-Blanc, Sylvain Univ Toulouse IRIT CNRS UMR5505 Toulouse France

ISBN: (纸本)9798400701207

The automatic construction of an image filter is a difficult task for which many recent machine-learning methods have been proposed. Cartesian Genetic Programming (CGP) has been effectively used in image-processing tasks by evolving programs with a function set specialized for computer vision. Although standard CGP is able to construct understandable image filter programs, we hypothesize that explicitly using a mechanism to control the size of the generated filter programs would help reduce the size of the final solution while keeping comparable efficacy on a given task. It is indeed central to keep the graph size as contained as possible as it improves our ability to understand them and explain their inner functioning. In this work, we use the Lexicase selection as the mechanism to control the size of the programs during the evolutionary process, by allowing CGP to evolve solutions based on performance and on the size of such solutions. We extend Kartezio, a Cartesian Genetic Programming for computer vision tasks, to generate our programs. We found in our preliminary experiment that CGP with Lexicase selection is able to achieve similar performance to the standard CGP while keeping the size of the solutions smaller.

关键词： evolutionary computation Cartesian Genetic Programming Lexicase Selection graph-based methods

来源：评论

学校读者我要写书评

暂无评论

The development of habitable urban skyways: claiming interstitial territories through evolutionary processes

引用

City, Territory and Architecture 2023年第1期10卷 1-19页

作者： Blewett, Liam Nguyen, Ngoc Xuan Huy Makki, Mohammed Showkatbakhsh, Milad University of Technology Sydney Sydney Australia Architectural Association London United Kingdom

Urban skyways, in which an elevated pedestrian-friendly layer of the city is applied to the existing urban fabric, have evolved from radical conceptual proposals in the mid-twentieth century, such as the continuous monument by SuperStudio, to increasingly realised proposals over the last 30 years, such as the highline project in Manhattan. Developed as either networks that respond to harsh climatic conditions, or solutions that aim to reclaim part of the city for pedestrian use, their use has been additive rather than integrative, in most cases culminating in elevated walkways. This article proposes an alternative approach to the utilisation of urban skyways within existing cities, in which the question of habitation is a primary driver. The research involved the application of sequential evolutionary simulations, to locations in the city of Paris, as an algorithmic approach to achieve multiple conflicting objectives. The results demonstrate the value of urban skyways as habitable spaces that respond to the city as well as being informed by it, in which issues of overpopulation, lack of public space, climatic response and urban farmlands—all of which impact most megacities today—shape the urban decision-making process. © 2023, Springer-Verlag GmbH Germany, part of Springer Nature.

关键词： evolutionary computation Habitable skyways Interstitial territory Paris Sequential evolutionary simulations

来源：评论

学校读者我要写书评

暂无评论

Generating Diverse and Discriminatory Knapsack Instances by Searching for Novelty in Variable Dimensions of Feature-Space 23

Generating Diverse and Discriminatory Knapsack Instances by ...

引用

Genetic and evolutionary computation Conference (GECCO)

作者： Marrero, Alejandro Segredo, Eduardo Hart, Emma Bossek, Jakob Neumann, Aneta Univ La Laguna San Cristobal La Laguna Spain Edinburgh Napier Univ Edinburgh Midlothian Scotland Rhein Westfal TH Aachen Aachen Germany Univ Adelaide Adelaide SA Australia

ISBN: (纸本)9798400701191

Generating new instances via evolutionary methods is commonly used to create newbenchmarking data-sets, with a focus on attempting to cover an instance-space as completely as possible. Recent approaches have exploited Quality-Diversity methods to evolve sets of instances that are both diverse and discriminatory with respect to a portfolio of solvers, but these methods can be challenging when attempting to find diversity in a high-dimensional feature-space. We address this issue by training a model based on Principal Component Analysis on existing instances to create a low-dimension projection of the high-dimension feature-vectors, and then apply Novelty Search directly in the new low-dimension space. We conduct experiments to evolve diverse and discriminatory instances of Knapsack Problems, comparing the use of Novelty Search in the original feature-space to using Novelty Search in a low-dimensional projection, and repeat over a given set of dimensions. We find that the methods are complementary: if treated as an ensemble, they collectively provide increased coverage of the space. Specifically, searching for novelty in a low-dimension space contributes 56% of the filled regions of the space, while searching directly in the feature-space covers the remaining 44%.

关键词： Instance generation instance-space analysis knapsack problem novelty search evolutionary computation

来源：评论

学校读者我要写书评

暂无评论

Choosing Representation, Mutation, and Crossover in Genetic Algorithms

引用

IEEE computationAL INTELLIGENCE MAGAZINE 2022年第4期17卷 52-53页

作者： Dockhorn, Alexander Lucas, Simon Leibniz Univ Hannover Hannover Germany Queen Mary Univ London London England

This paper aims to provide an introduction to genetic algorithms and their three main components, i.e., the representation of solutions and their modification through mutation and crossover operators. It has been specifically designed as introduction for newcomers to this exciting research area. This short paper represents a summary of the full paper found online in IEEE Xplore. The latter provides interactive components for a hands-on exploration of the covered material.

关键词： evolutionary computation Genetics Genetic algorithms

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：