Contemporary Symbolic Regression Methods and their Relative Performance
arXiv:2107.14351
Abstract
Many promising approaches to symbolic regression have been presented in recent years, yet progress in the field continues to suffer from a lack of uniform, robust, and transparent benchmarking standards. In this paper, we address this shortcoming by introducing an open-source, reproducible benchmarking platform for symbolic regression. We assess 14 symbolic regression methods and 7 machine learning methods on a set of 252 diverse regression problems. Our assessment includes both real-world datasets with no known model form as well as ground-truth benchmark problems, including physics equations and systems of ordinary differential equations. For the real-world datasets, we benchmark the ability of each method to learn models with low error and low complexity relative to state-of-the-art machine learning methods. For the synthetic problems, we assess each method's ability to find exact solutions in the presence of varying levels of noise. Under these controlled experiments, we conclude that the best performing methods for real-world regression combine genetic algorithms with parameter estimation and/or semantic search drivers. When tasked with recovering exact equations in the presence of noise, we find that deep learning and genetic algorithm-based approaches perform similarly. We provide a detailed guide to reproducing this experiment and contributing new methods, and encourage other researchers to collaborate with us on a common and living symbolic regression benchmark.
arXiv commentTo appear in Neurips 2021 Track on Datasets and Benchmarks. Main text: 10 pages, 3 figures; Appendix: 7 pages, 8 figures. https://openreview.net/forum?id=xVQMrDLyGst
References in corpus (10)
- Scikit-learn: Machine Learning in Python
- Artificial Intelligence: the global landscape of ethics guidelines
- Interpretable machine learning: definitions, methods, and applications
- OpenML: networked science in machine learning
- Preventing Fairness Gerrymandering: Auditing and Learning for Subgroup Fairness
- AI Feynman 2.0: Pareto-optimal symbolic regression exploiting graph modularity
- A Greedy Search Tree Heuristic for Symbolic Regression
- Bayesian Symbolic Regression
- Learning concise representations for regression by evolving networks of trees
- Informed Equation Learning
Cited by in corpus (18)
- A computational framework for physics-informed symbolic regression with straightforward integration of domain knowledge
- Exhaustive Symbolic Regression
- Discovering interpretable elastoplasticity models via the neural polynomial method enabled symbolic regressions
- On the functional form of the radial acceleration relation
- A New Deterministic Technique for Symbolic Regression
- Symbolic Regression as Feature Engineering Method for Machine and Deep Learning Regression Tasks
- Cosmic Inflation and Genetic Algorithms
- Automated Learning of Interpretable Models with Quantified Uncertainty
- MMSR: Symbolic Regression is a Multi-Modal Information Fusion Task
- TorchSISSO: A PyTorch-Based Implementation of the Sure Independence Screening and Sparsifying Operator for Efficient and Interpretable Model Discovery
- SymbolNet: Neural Symbolic Regression with Adaptive Dynamic Pruning for Compression
- Transformation-Interaction-Rational Representation for Symbolic Regression
- Information Fusion via Symbolic Regression: A Tutorial in the Context of Human Health
- cp3-bench: A tool for benchmarking symbolic regression algorithms tested with cosmology
- SymbolFit: Automatic Parametric Modeling with Symbolic Regression
- Simultaneous Model-Based Evolution of Constants and Expression Structure in GP-GOMEA for Symbolic Regression
- Dimension Reduction for Symbolic Regression
- Untangling the Effects of Down-Sampling and Selection in Genetic Programming