This book brings together some of the most impactful researchers in the field of Genetic Programming (GP), each one working on unique and interesting intersections of theoretical development and practical applications of this evolutionary-based Machine Learning paradigm. Topics of particular interest for this year´s book include powerful modeling techniques through GP-based symbolic regression, novel selection mechanisms that help guide the evolutionary process, modular approaches to GP, and applications in cybersecurity, biomedicine and program synthesis, as well as papers by practitioner of GP that focus on usability and real-world results. In summary, readers will get a glimpse of the current state of the art in GP research.
We present a GUI-driven and efficient Genetic Programming (GP) and AI Planning framework designed for agent-based learning research. Our framework, ABL-Unity3D, is built in Unity3D, a game development environment. ABL-Unity3D addresses challenges entailed in co-opting Unity3D: making the simulator serve agent learning rather than humans playing a game, lowering fitness evaluation time to make learning computationally feasible, and interfacing GP with an AI Planner to support hybrid algorithms. We achieve this by developing a Graphical User Interface (GUI) using the Unity3D editor’s programmable interface and performance optimizations. These optimizations result in at least a 3x speedup. In addition, we describe ABL-Unity3D by explaining how to use it for an example experiment using GP and AI Planning. We benchmark ABL-Unity3D by measuring the performance and speed of the AI Planner alone, GP alone, and the AI Planner with GP.
Increasing demand for human understanding of machine decision-making is deemed crucial for machine learning (ML) methodology development and further applications. It has inspired the emerging research field of interpretable and explainable ML/AI. Techniques have been developed to either provide additional explanations to a trained ML model or learn innately compact and understandable models. Genetic programming (GP), as a powerful learning instrument, holds great potential in interpretable and explainable learning. In this chapter, we first discuss concepts and popular methods in interpretable and explainable ML, and review research using GP for interpretability and explainability. We then introduce our previously proposed GP-based framework for interpretable and explainable learning applied to bioinformatics.
Author(s): Leonardo Trujillo ,Stephan M. Winkler, Sara Silva, Wolfgang Banzhaf
Publisher: Springer
Year: 2023
Language: English
Pages: 272
978-981-19-8460-0
1
Preface
Acknowledgements
Contents
Contributors
978-981-19-8460-0_1
Symbolic Regression in Materials Science: Discovering Interatomic Potentials from Data
1 Introduction
1.1 Materials Informatics and Data-Driven Potentials
1.2 Current Challenges
2 State of the Art
2.1 Directed Search
2.2 Directed Search with Parallel Multilevel Genetic Program
2.3 Parallel Tempering
2.4 Symbolically Regressed Table KMC
2.5 Hierarchical Fair Competition
2.6 Potential Optimization by Evolutionary Techniques (POET)
2.7 Other Applications
2.8 Summary Discussion
3 Designing GP for Modeling Interatomic Potentials
3.1 Symbolic Regression in Operon
3.2 Empirical Validation
4 Conclusion
5 Appendix
References
978-981-19-8460-0_2
Correlation Versus RMSE Loss Functions in Symbolic Regression Tasks
1 Introduction
2 Related Work
3 Benchmarks
3.1 Koza's Benchmarks
3.2 New Benchmark Standards
3.3 The GPTP Benchmarks
3.4 Feynman Symbolic Regression Benchmark
4 Methods
4.1 Noise Introduction
4.2 Varying Number of Points
4.3 Dimensional Sensitivity
4.4 Sensitivity to Constants
5 Results
5.1 The Keijzer-5 Benchmark
5.2 The Korns-8 Benchmark
5.3 The Vladislavleva-1 Benchmark
5.4 The Nguyen-5 Benchmark
5.5 Feynman Symbolic Regression Benchmark
5.6 Sensitivity to Constants
6 Discussion
7 Conclusions
8 Appendix
References
978-981-19-8460-0_3
GUI-Based, Efficient Genetic Programming and AI Planning for Unity3D
1 Introduction
2 ABL-Unity3D
2.1 Simulator
2.2 Genetic Programming
2.3 AI Planning
2.4 Graphical User Interface (GUI)
2.5 Optimizations
3 ABL-Unity3D Example Scenario: N Prongs
3.1 Scenario Specification
3.2 Solution Specification
3.3 Benchmark Performance
4 Discussion
5 Future Work
References
978-981-19-8460-0_4
Genetic Programming for Interpretable and Explainable Machine Learning
1 Introduction
2 Background
2.1 Explainability and Interpretability
2.2 Genetic Programming for Explainable and Interpretable Learning
3 An Explainable and Interpretable Learning Algorithm for Metabolomics
3.1 Method and Data
3.2 Web Interface
3.3 Results
4 Conclusion
References
978-981-19-8460-0_5
Biological Strategies ParetoGP Enables Analysis of Wide and Ill-Conditioned Data from Nonlinear Systems
1 Goals and Objectives
2 Illustration Data Sets
2.1 MSC Metabolites
2.2 CAR T-Cell
2.3 Bone Regeneration Biomarkers
2.4 Single Cell Multiomics
3 The Application Space
3.1 Biological Data
3.2 Definitions of Success
3.3 Data Challenges
3.4 Biological Data Trends
4 ParetoGP Foundations
4.1 Essential Assumptions
4.2 Evolutionary Basics
4.3 Modeling Nuances
4.4 Ensembles Enable Trustable Models
5 The Analysis Workflow
5.1 The Role of the Data Owner
5.2 Model Evolution
5.3 Interesting Models—The Knee of the Pareto Front
5.4 Variable Selection—Making the Haystack Smaller
5.5 MetaVariables and Basis Functions
5.6 Model Selection and Ensemble Definition
5.7 Data Cubes and Summary Statistics
5.8 Trade-Off Analysis
6 Competing Technologies
7 Conclusions
8 Research Opportunities
References
978-981-19-8460-0_6
GP-Based Generative Adversarial Models
1 Introduction
2 State of the Art
3 TensorGP
3.1 Genotype to Phenotype Mapping
3.2 Function and Terminal Set
3.3 Speed Gains
4 GP-Based ANN-Guided Image Generation
4.1 Experimental Results
5 Framework—TGPGAN
6 GP-Based Adversarial Image Generation
6.1 Experimental Results
7 Conclusions and Future Work
References
978-981-19-8460-0_7
Modeling Hierarchical Architectures with Genetic Programming and Neuroscience Knowledge for Image Classification Through Inferential Knowledge
1 Introduction
2 Myths and Prospects of Genetic Programming
3 Inferential Knowledge
3.1 Deductive Reasoning
3.2 Inductive Reasoning
3.3 Abductive Reasoning
3.4 Retroductive Reasoning
3.5 Transductive Reasoning
4 Inferential Knowledge in Brain Programming
4.1 Doublets
4.2 The Visual Turing Test
4.3 Abductive Reasoning in Brain Programming
5 Results
6 Conclusions
References
978-981-19-8460-0_8
Life as a Cyber-Bio-Physical System
1 Introduction
2 The Triangle of Life
2.1 B: Biology
2.2 P: Physics
2.3 C: Computing
2.4 B-P: Bio-physical Systems
2.5 C-B: Cyber-Bio Systems
2.6 C-P: Cyber-Physical Systems
2.7 Pulling It All Together: C-B-P
3 Zoetic Science
3.1 Philosophy
3.2 Systems
3.3 Mathematics
3.4 Simulation and In Silico Models
3.5 Scale, Complexity, and Emergence
3.6 Far-From-Equilibrium Thermodynamics
3.7 Embodiment v. Virtual Physics
3.8 Metamaterials
3.9 Meta-Dynamics
4 Zoetic Engineering
5 Conclusions
References
978-981-19-8460-0_9
STREAMLINE: A Simple, Transparent, End-To-End Automated Machine Learning Pipeline Facilitating Data Analysis and Algorithm Comparison
1 Introduction
2 Methods
2.1 STREAMLINE
2.2 Evaluation Datasets
3 Results and Discussion
3.1 HCC UCI Benchmark Data Results
3.2 GAMETES Simulated Datasets Results
3.3 x-bit MUX Benchmark Datasets Results
4 Conclusions
References
978-981-19-8460-0_10
Evolving Complexity is Hard
1 Introduction
2 The Digital Circuit G-P Map
2.1 Neutral and Epochal Evolution
3 Genotype-Phenotype Maps
3.1 Phenotype Network and Neutral Sets
3.2 Redundancy
3.3 Robustness
3.4 Evolvability
3.5 Universal Structural Properties
4 Structural and Complexity Properties for the Circuit G-P Map
4.1 Redundancy and Bias
4.2 Robustness
4.3 Evolvability
4.4 Tononi and Kolmogorov Complexity
4.5 Complexity Density
4.6 Complexity Correlations
4.7 Complexity and Redundancy
4.8 Complexity and Robustness
4.9 Complexity and Evolvability
5 Conclusion
5.1 Further Work
References
978-981-19-8460-0_11
ESSAY: Computers Are Useless ... They Only Give Us Answers
1 Introduction
2 Recommended Sources
2.1 Hidden Order
2.2 Gödel, Escher, Bach
2.3 The Cambridge Quintet
3 Artificial Intelligence
4 The Mangle of Practice, Times, Agency, and Science
5 Combinators
6 Where Next?
7 Summary
References
1 (1)
Index
