GP - Genetic Programming
Description
Genetic Programming (GP) is an evolutionary computation technique that automatically generates computer programs to solve specific problems. The search is guided by a fitness function based on input-output examples, usually without further human intervention. Different program representations have been used in GP to help traverse the search space, such as tree structures, linear sequences of code, graphs, and grammars. More recently, different variants of hybridizsing large language models (LLMs) with evolutionary search have been proposed to add context additionally to input-output examples. The GP track invites original contributions on all aspects of evolutionary generation of computer programs or other executable structures for specific tasks.
Scope
Advances in genetic programming include but are not limited to:
- Analysis: Information Theory, Complexity, Run-time, Visualization, Fitness Landscape, Bloat, Generalisation, Domain adaptation, Security, Alignment, Innovation, Creativity Applications: Computer Science, Mathematics, Engineering, Physics, Biology, Medicine, Finance, Economics, AI
- Environments: Static, Dynamic, Interactive, Uncertain, Online, Offline, Simulated, Robotic
- Operators: Selection, Crossover, Mutation, Variation, Replacement, LLM-based
- Performance: Surrogate functions, Multi-Objective, Coevolutionary, Human Competitive, Caching, Semantic duplicates, Diversity, Inheritance, Tracing
- Populations: Islands, Demes, Niches, MAP-Elites, Parallel, Distributed, Agents
- Programs: Decomposition, Modularity, Semantics, Simplification, Software Improvement, Bug Repair, Software/Program Testing
- Representations: Cartesian, Grammatical, Graphs, Linear, Rules, Trees, Geometric and Semantic
- Synthesis: Programs, Algorithms, Circuits, Systems.
Track Chairs
Gabriel Kronberger
University of Applied Sciences Upper Austria | webpage
Gabriel Kronberger is a Professor at the University of Applied Sciences Upper Austria and has been working on tree-based genetic programming methods especially for symbolic regression since 20 years. His research interests are symbolic regression for physics-based machine learning and applications in science and engineering.Fabricio Olivetti de França
Federal University of ABC (UFABC), Brazil
Fabricio Olivetti de Franca: is a Professor of Computer Science at the Federal University of ABC, current head of the Heuristics, Analysis and Learning Laboratory (HAL) and the coordinator of the graduate program of computer science at the same university. His work in symbolic regression comprehends the creation of new techniques promoting interpretability and the integration of domain knowledge. He is also one of the main contributors of SRBench having helped to host multiple competitions for symbolic regression and the current version of SRBench. He co-organized Symbolic Regression workshops at GECCO for the past years together with Gabriel Kronberger and William La Cava. He is also part of the organization of the Genetic Programming Theory and Practice workshop.