The Opaque Nature of Intelligence and the Pursuit of Explainable AI DOI Creative Commons
Sarah Thomson, Bas van Stein, Daan van den Berg

et al.

Published: Jan. 1, 2023

In this work We consider and discuss the problems which come with trying to explain human machine intelligence.How explainable artificial intelligence research is being carried out, pitfalls limitations of current approaches bigger question whether we need explanations for trusting inherently complex large intelligent systems, or not.

Language: Английский

The intersection of evolutionary computation and explainable AI DOI
Jaume Bacardit, Alexander E. I. Brownlee, Stefano Cagnoni

et al.

Proceedings of the Genetic and Evolutionary Computation Conference Companion, Journal Year: 2022, Volume and Issue: unknown, P. 1757 - 1762

Published: July 9, 2022

In the past decade, Explainable Artificial Intelligence (XAI) has attracted a great interest in research community, motivated by need for explanations critical AI applications. Some recent advances XAI are based on Evolutionary Computation (EC) techniques, such as Genetic Programming. We call this trend EC XAI. argue that full potential of methods not been fully exploited yet XAI, and community future efforts field. Likewise, we find there is growing concern regarding explanation population-based methods, i.e., their search process outcomes. While some attempts have done direction (although, most cases, those explicitly put context XAI), believe still several opportunities open questions that, principle, may promote safer broader adoption real-world within EC. position paper, briefly overview main results two above trends, suggest play major role achievement

Language: Английский

Citations

36
XAI for Algorithm Configuration and Selection DOI
Sarah L. Thomson, Emma Hart, Quentin Renau

et al.

Natural computing series, Journal Year: 2025, Volume and Issue: unknown, P. 117 - 148

Published: Jan. 1, 2025

Language: Английский

Citations

0
The Synergy of Explainable AI and Evolutionary Computation: Real-World Applications DOI
Bas van Stein, Qi Huang, Elena Raponi

et al.

Natural computing series, Journal Year: 2025, Volume and Issue: unknown, P. 175 - 195

Published: Jan. 1, 2025

Language: Английский

Citations

0
Exploratory Landscape Analysis DOI Open Access
Pascal Kerschke, Mike Preuß

Published: July 15, 2023

tutorial Free Access Share on Exploratory Landscape Analysis Authors: Pascal Kerschke "Friedrich List" Faculty of Transport and Traffic Sciences, TU Dresden, Germany Center for Scalable Data Science (ScaDS.AI), Dresden/Leipzig, https://orcid.org/0000-0003-2862-1418Search about this author , Mike Preuss LIACS, University Leiden, Netherlands https://orcid.org/0000-0003-4681-1346Search Authors Info & Claims GECCO '23 Companion: Proceedings the Companion Conference Genetic Evolutionary ComputationJuly 2023Pages 990–1007https://doi.org/10.1145/3583133.3595058Published:24 July 2023Publication History 0citation0DownloadsMetricsTotal Citations0Total Downloads0Last 12 Months0Last 6 weeks0 Get Citation AlertsNew Alert added!This alert has been successfully added will be sent to:You notified whenever a record that you have chosen cited.To manage your preferences, click button below.Manage my Alert!Please log in to account Save BinderSave BinderCreate New BinderNameCancelCreateExport CitationPublisher SiteeReaderPDF

Language: Английский

Citations

6
Towards explainable metaheuristics: Feature extraction from trajectory mining DOI Creative Commons
Martin Fyvie, John McCall, Lee A. Christie

et al.

Expert Systems, Journal Year: 2023, Volume and Issue: unknown

Published: Nov. 2, 2023

Abstract Explaining the decisions made by population‐based metaheuristics can often be considered difficult due to stochastic nature of mechanisms employed these optimisation methods. As industries continue adopt methods in areas that increasingly require end‐user input and confirmation, need explain internal being has grown. In this article, we present our approach extraction explanation supporting features using trajectory mining. This is achieved through application principal components analysis techniques identify new tracking population diversity changes post‐runtime. The algorithm search trajectories were generated solving a set benchmark problems with genetic univariate estimation distribution retaining all visited candidate solutions which then projected lower dimensional sub‐space. We also varied selection pressure placed on high fitness altering operators. Our results show metrics derived from sub‐space are capable capturing key learning steps how solution variable patterns function may captured component coefficients. A comparative study importance rankings surrogate model built same dataset was performed. both approaches identifying regarding interactions their influence complimentary fashion.

Language: Английский

Citations

4
On the Latent Structure of the bbob-biobj Test Suite DOI
Pavel Krömer, Vojtéch Uher, Tea Tušar

et al.

Lecture notes in computer science, Journal Year: 2024, Volume and Issue: unknown, P. 326 - 341

Published: Jan. 1, 2024

Language: Английский

Citations

0
Fitness Landscape k-Nearest Neighbors Classification Based on Fitness Values Distribution DOI
Vojtéch Uher, Pavel Krömer

2022 IEEE Congress on Evolutionary Computation (CEC), Journal Year: 2024, Volume and Issue: unknown, P. 1 - 9

Published: June 30, 2024

Language: Английский

Citations

0
Towards an Improved Understanding of Features for More Interpretable Landscape Analysis DOI
Marcus Gallagher, Mario Andrés Muñoz

Proceedings of the Genetic and Evolutionary Computation Conference Companion, Journal Year: 2024, Volume and Issue: unknown, P. 135 - 138

Published: July 14, 2024

Language: Английский

Citations

0
Algorithm Instance Footprint: Separating Easily Solvable and Challenging Problem Instances DOI Creative Commons
Ana Nikolikj, Sašo Džeroski, Mario Andrés Muñoz

et al.

arXiv (Cornell University), Journal Year: 2023, Volume and Issue: unknown

Published: Jan. 1, 2023

In black-box optimization, it is essential to understand why an algorithm instance works on a set of problem instances while failing others and provide explanations its behavior. We propose methodology for formulating footprint that consists are easy be solved difficult solved, instance. This behavior the further linked landscape properties which make some or challenging. The proposed uses meta-representations embed performance into same vector space. These obtained by training supervised machine learning regression model prediction applying explainability techniques assess importance features predictions. Next, deterministic clustering demonstrates using them captures across space detects regions poor good performance, together with explanation leading it.

Language: Английский

Citations

0
From Selecting Best Algorithm to Explaining Why It is: A General Review, Formal Problem Statement and Guidelines Towards to an Empirical Generalization DOI
Vanesa Landero N., Joaquín Pérez-Ortega,

Carlos Andrés Collazos Morales

et al.

Lecture notes in computer science, Journal Year: 2023, Volume and Issue: unknown, P. 694 - 712

Published: Jan. 1, 2023

Language: Английский

Citations

0