Symbolic regression is a common application forgenetic programming (GP). we present a newnon-evolutionary technique for symbolic regressionthat, compared to competent GP approaches on real-worldproblems, is orders of magnitude faster (taking justseconds), returns simpler models, has comparable orbetter prediction on unseen data, and convergesreliably and deterministically. I dub the approach FFX,for Fast Function Extraction. FFX uses a recentlydeveloped machine learning technique, pathwiseregularised learning, to rapidly prune a huge set ofcandidate basis functions down to compact models. FFXis verified on a broad set of real-world problemshaving 13 to 1468 input variables, outperforming GP aswell as several state-of-the-art regressiontechniques.
CITATION STYLE
McConaghy, T. (2011). FFX: Fast, Scalable, Deterministic Symbolic Regression Technology (pp. 235–260). https://doi.org/10.1007/978-1-4614-1770-5_13
Mendeley helps you to discover research relevant for your work.