Genetic Programming : 20th European Conference, EuroGP 2017, Amsterdam, The Netherlands, April 19-21, 2017, Proceedings / / edited by James McDermott, Mauro Castelli, Lukas Sekanina, Evert Haasdijk, Pablo García-Sánchez

• Pubblicazione: Cham : , : Springer International Publishing : , : Imprint : Springer, , 2017
• Edizione: 1st ed. 2017.
• Descrizione fisica: 1 online resource (XII, 359 p. 116 illus.)
• Disciplina: 006.31
• Soggetto topico: Algorithms; Computer science; Artificial intelligence; Application software; Computer science - Mathematics; Discrete mathematics; Theory of Computation; Artificial Intelligence; Computer and Information Systems Applications; Discrete Mathematics in Computer Science
• Persona (resp. second.): McDermottJames; CastelliMauro; SekaninaLukas; HaasdijkEvert; García-SánchezPablo
• Nota di contenuto: Intro -- Preface -- Organization -- Contents -- Oral Presentations -- Evolutionary Program Sketching -- 1 Introduction -- 2 Program Sketching -- 3 Evolutionary Program Sketching -- 3.1 Problem Specification -- 3.2 Instruction Set -- 3.3 Fitness Function -- 3.4 Exploiting the Feedback from Hole Completion -- 4 Related Work -- 5 Experimental Evaluation -- 6 Discussion -- 7 Conclusion -- References -- Exploring Fitness and Edit Distance of Mutated Python Programs -- 1 Introduction -- 2 Related Work -- 3 Our Implementation of Genetic Improvement -- 3.1 Fitness Function -- 3.2 Search Algorithm -- 4 Experimental Setup -- 4.1 Description of the Programs Targeted by GI -- 5 Results -- 5.1 Change in Fitness -- 5.2 Average Fitness with Respect to Edit List Size -- 5.3 Discrete Steps in Fitness -- 6 Conclusions -- References -- Differentiable Genetic Programming -- 1 Introduction -- 2 Program Encoding -- 3 The Algebra of Truncated Polynomials -- 3.1 The Link to Taylor Polynomials -- 3.2 Non Rational Functions -- 4 Example of a dCGP -- 5 Learning Constants in Symbolic Regression -- 5.1 Ephemeral Constants Approach -- 5.2 Weighted dCGP Approach -- 6 Solution to Differential Equations -- 7 Discovery of Prime Integrals -- 8 Conclusions -- References -- Evolving Game State Features from Raw Pixels -- 1 Introduction -- 2 Related Research -- 3 Materials -- 3.1 Games -- 3.2 Handcrafted Game State Features -- 4 Evolving Video Game State Visual Features Using Genetic Programming -- 4.1 Evolving Game State Features -- 4.2 Voting for Actions -- 5 Results -- 6 Conclusion -- References -- Emergent Tangled Graph Representations for Atari Game Playing Agents -- 1 Introduction -- 2 Background -- 3 The Arcade Learning Environment -- 3.1 Screen State Space Representation -- 4 Evolving Tangled Program Graphs -- 4.1 Coevolving Teams of Programs -- 4.2 Emergent Modularity. ; 4.3 Diversity Maintenance -- 5 Empirical Experiments -- 5.1 Experimental Setup -- 5.2 Results -- 5.3 Solution Analysis -- 6 Conclusion and Future Work -- References -- A General Feature Engineering Wrapper for Machine Learning Using -Lexicase Survival -- 1 Introduction -- 2 Feature Engineering Wrapper -- 2.1 -lexicase Survival -- 2.2 Scaling -- 3 Related Work -- 4 Experimental Analysis -- 4.1 Problems -- 5 Results -- 5.1 Hyper-Parameter Optimization -- 5.2 Problem Performance -- 5.3 Statistical Analysis -- 6 Discussion -- 7 Conclusions -- References -- Visualising the Search Landscape of the Triangle Program -- 1 Genetic Improvement -- 2 Triangle Program Software Engineering Benchmark -- 3 Binary Representation: Replacing Comparisons with One Alternative -- 3.1 High Order Binary Schema Are Not Deceptive -- 3.2 Binary Schema Predict All Solutions of the Triangle Program -- 3.3 Local Search Landscape of the Binary Space -- 4 Original All Comparisons -- 4.1 Fitness Space of Triangle Program -- 4.2 High Order Schema Analysis -- 4.3 Local Search for the Triangle Program -- 4.4 Local Optima Networks -- 5 Conclusions -- References -- RANSAC-GP: Dealing with Outliers in Symbolic Regression with Genetic Programming -- 1 Introduction -- 2 Background -- 2.1 Outliers -- 3 Robust Regression -- 4 Proposed RANSAC-GP -- 4.1 Proposal -- 5 Experiments and Results -- 5.1 Results -- 6 Conclusion and Future Work -- References -- Symbolic Regression on Network Properties -- 1 Introduction -- 2 Related Work -- 2.1 Symbolic Regression -- 2.2 Cartesian Genetic Programming (CGP) -- 3 Networks -- 3.1 Network Representations -- 3.2 Network Properties -- 4 Experiments -- 4.1 Network Diameter -- 4.2 Isoperimetric Number -- 5 Discussion -- 5.1 Network Diameter -- 5.2 Isoperimetric Number -- 6 Conclusion -- References. ; Evolving Time-Invariant Dispatching Rules in Job Shop Scheduling with Genetic Programming -- 1 Introduction -- 1.1 Goals -- 1.2 Organisation -- 2 Background -- 2.1 Job Shop Scheduling -- 2.2 Automatic Design of Dispatching Rules -- 3 Time-Invariant Dispatching Rule -- 3.1 An Example: Time-Invariance v.s. Time-Dependence -- 3.2 Relationship Between Existing Rule Classifications -- 4 Selection of Terminals for Time-Invariance -- 5 Experimental Studies -- 5.1 Results and Discussions -- 5.2 Further Analysis -- 5.3 Time-Invariance of the Evolved Rules -- 6 Conclusions and Future Work -- References -- Strategies for Improving the Distribution of Random Function Outputs in GSGP -- 1 Introduction -- 2 Background and Motivation -- 2.1 Geometric Semantic Genetic Programming -- 2.2 The Impact of the Random Functions -- 3 Related Work -- 4 Strategies for Normalizing Outputs of Random Functions -- 5 Experimental Analysis -- 5.1 Normalization Impact on the Distribution of the Semantics of Random Functions -- 5.2 The Impact on the GSGP Performance -- 6 Conclusions and Future Work -- References -- Synthesis of Mathematical Programming Constraints with Genetic Programming -- 1 Introduction -- 2 Related Work -- 3 Constraint Synthesis -- 3.1 Constraint Synthesis Problem -- 3.2 Genetic Constraint Synthesis (GenetiCS) -- 4 Experiment -- 4.1 Setup -- 4.2 Evaluation of GP Setups -- 4.3 Evaluation of Synthesized Models -- 5 Conclusion -- References -- Grammatical Evolution of Robust Controller Structures Using Wilson Scoring and Criticality Ranking -- 1 Introduction -- 2 Background -- 2.1 Grammatical Evolution -- 2.2 Robust Control -- 3 Methodology -- 3.1 General Process -- 3.2 Wilson Scoring -- 3.3 Criticality Ranking -- 4 Experiments -- 4.1 Benchmark Problem -- 4.2 Metrics and Setup -- 4.3 Results -- 5 Conclusion -- References. ; Using Feature Clustering for GP-Based Feature Construction on High-Dimensional Data -- 1 Introduction -- 2 Background -- 2.1 Genetic Programming for Feature Construction -- 2.2 Feature Clustering -- 3 The Proposed Approach -- 3.1 The Redundancy Based Feature Clustering Method: RFC -- 3.2 The Proposed Method: CGPFC -- 4 Experiment Design -- 5 Results and Discussions -- 5.1 Performance of the Constructed Feature -- 5.2 Performance of the Constructed and Selected Features -- 5.3 Cluster Analysis -- 6 Conclusions and Future Work -- References -- Posters -- Geometric Semantic Crossover with an Angle-Aware Mating Scheme in Genetic Programming for Symbolic Regression -- 1 Introduction -- 2 Background -- 2.1 Geometric Semantic GP -- 2.2 Locally Geometric Semantic Crossover -- 2.3 Related Work -- 3 Angle-Aware Geometric Semantic Crossover (AGSX) -- 3.1 Main Idea -- 3.2 The AGSX Process -- 3.3 Main Characteristics of AGSX -- 3.4 Fitness Function of the Algorithm -- 4 Experiments Setup -- 4.1 Benchmark Problems -- 4.2 Parameter Settings -- 5 Results and Discussions -- 5.1 Overall Results -- 5.2 Analysis on the Learning Performance -- 5.3 Analysis of the Evolution of Generalisation Performance -- 5.4 Analysis of the Angles -- 5.5 Comparison on Computational Time and Program Size -- 6 Conclusions and Future Work -- References -- RECIPE: A Grammar-Based Framework for Automatically Evolving Classification Pipelines -- 1 Introduction -- 2 Related Work -- 3 Automatically Evolving Classification Pipelines -- 3.1 Grammar: Representing Effective Classification Pipelines -- 3.2 Individual Representation -- 3.3 Individual Evaluation -- 4 Experimental Results -- 4.1 Comparison with Other State-of-the-Art Methods -- 4.2 Analysis of the Evolutionary Process of RECIPE -- 5 Conclusions and Future Work -- References. ; A Grammar Design Pattern for Arbitrary Program Synthesis Problems in Genetic Programming -- 1 Introduction -- 2 System Description -- 2.1 Grammar -- 2.2 Skeleton -- 2.3 Python Specific Differences -- 2.4 Implementation Details -- 3 Previous Approaches to Program Synthesis -- 3.1 PushGP -- 3.2 Strongly Formed Genetic Programming -- 3.3 Grammar Guided Genetic Programming -- 3.4 Program Synthesis via Code Reusage -- 3.5 Comparison of Program Synthesis Approaches -- 4 Experimental Setup -- 4.1 Benchmark Suite -- 4.2 Experimental Parameter Settings -- 4.3 PushGP Differences -- 5 Results -- 5.1 Comparison to PushGP on Tournament Selection -- 5.2 Comparison to PushGP on Lexicase Selection -- 5.3 Generational Progress -- 5.4 Invalids -- 6 Conclusion and Future Work -- References -- Improving the Tartarus Problem as a Benchmark in Genetic Programming -- 1 Introduction -- 2 Desirable GP Benchmark Characteristics -- 3 GP Benchmarks -- 3.1 The Lawnmower Problem -- 4 The Tartarus Problem -- 4.1 Satisfying the Desirable Benchmark Characteristics -- 4.2 Current State Evaluation -- 4.3 Proposed Improved State Evaluation -- 4.4 Baseline Values for Tartarus Instances -- 4.5 Generating Tartarus Instances -- 4.6 Tuning Difficulty -- 5 Conclusion -- References -- A New Subgraph Crossover for Cartesian Genetic Programming -- 1 Introduction -- 2 Related Work -- 2.1 Cartesian Genetic Programming -- 2.2 Previous Work on Crossover in CGP -- 3 The Proposed Method -- 3.1 Multiple Outputs -- 4 Experiments -- 4.1 Experimental Setup -- 4.2 Symbolic Regression -- 4.3 Boolean Functions -- 4.4 Image Operator Design -- 4.5 Crossover Comparison -- 5 Discussion -- 6 Conclusion and Future Work -- References -- A Comparative Study of Different Grammar-Based Genetic Programming Approaches -- 1 Introduction -- 2 Grammar-Based Genetic Programming. ; 2.1 Contex-Free-Grammar Genetic Programming (CFG-GP).
• Sommario/riassunto: This book constitutes the refereed proceedings of the 20th European Conference on Genetic Programming, EuroGP 2017, held in Amsterdam, The Netherlands, in April 2017, co-located with the Evo* 2017 events, EvoCOP, EvoMUSART, and EvoApplications. The 14 revised full papers presented together with 8 poster papers were carefully reviewed and selected from 32 submissions. The wide range of topics in this volume reflects the current state of research in the field. Thus, we see topics and applications including program synthesis, genetic improvement, grammatical representations, self-adaptation, multi-objective optimisation, program semantics, search landscapes, mathematical programming, games, operations research, networks, evolvable hardware, and program synthesis benchmarks.
• Titolo autorizzato: Genetic Programming
• ISBN: 3-319-55696-7
• Formato: Materiale a stampa
• Livello bibliografico: Monografia
• Lingua di pubblicazione: Inglese
• Record Nr.: 9910483925703321
• Lo trovi qui: Univ. Federico II
• Serie: Theoretical Computer Science and General Issues, . 2512-2029 ; ; 10196