Balancing Learning and Overfitting in Genetic Programming with Interleaved Sampling of Training Data

Gonçalves, Ivo; Silva, Sara

doi:10.1007/978-3-642-37207-0_7

Cited by 60 publications

(43 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This dataset is composed by 234 instances where each is a vector of of 627 elements (626 features that identify a drug, followed by the known LD50 value for that drug). For a more detailed description of the datasets please refer to Archetti et al [1] and Gonçalves and Silva [11]. The experiments described in this section correspond to a typical machine learning task and the datasets are divided in two equal parts.…”

Section: Predictive Modelingmentioning

confidence: 99%

“…Also, the solutions obtained in the last training stages are not clearly better than those discovered in the beginning, suggesting that the GE variants are not able to evolve reliable and general models. The LD50 dataset is considered as particularly challenging and standard GP-based approaches are unable to learn Genet Program Evolvable Mach models with good generalization ability [11]. The solution to overcome difficulties is to rely on specific training strategies that help to control overfitting.…”

Section: Predictive Modelingmentioning

confidence: 99%

See 1 more Smart Citation

Unveiling the properties of structured grammatical evolution

Lourenço

Pereira

Costa

2016

Genet Program Evolvable Mach

View full text Add to dashboard Cite

Structured grammatical evolution (SGE) is a new genotypic representation for grammatical evolution (GE). It comprises a hierarchical organization of the genes, where each locus is explicitly linked to a non-terminal of the grammar being used. This one-to-one correspondence ensures that the modification of a gene does not affect the derivation options of other non-terminals. We present a comprehensive set of optimization results obtained with problems from three different categories: symbolic regression, path finding, and predictive modeling. In most of the situations SGE outperforms standard GE, confirming the effectiveness of the new representation. To understand the reasons for SGE enhanced performance, we scrutinize its main features. We rely on a set of static measures to model the interactions between the representation and variation operators and assess how they influence the interplay between the genotype-phenotype spaces. The study reveals that the structured organization of SGE promotes an increased locality and is less redundant than standard GE, thus fostering an effective exploration of the search space.

show abstract

Section: Predictive Modelingmentioning

confidence: 99%

Section: Predictive Modelingmentioning

confidence: 99%

Unveiling the properties of structured grammatical evolution

Lourenço

Pereira

Costa

2016

Genet Program Evolvable Mach

View full text Add to dashboard Cite

show abstract

“…Selecting too many centroids can rob the validation and testing data sets of poorly represented data (e.g., large T mean , U sig , d 0 , d 50 ) and may tend to cause the GP to produce overly complex predictors (e.g., Gonçalves and Silva, 2013;Jensen, 1997, 1998). The selection of too few centroids can leave the testing data with too few data points to capture the variability in the data set (Goldstein et al, 2013).…”

Section: Selection Of Training Validation and Testing Data Setsmentioning

confidence: 99%

Data-driven components in a model of inner-shelf sorted bedforms: a new hybrid model

et al. 2014

View full text Add to dashboard Cite

Abstract. Numerical models rely on the parameterization of processes that often lack a deterministic description. In this contribution we demonstrate the applicability of using machine learning, a class of optimization tools from the discipline of computer science, to develop parameterizations when extensive data sets exist. We develop a new predictor for near-bed suspended sediment reference concentration under unbroken waves using genetic programming, a machine learning technique. We demonstrate that this newly developed parameterization performs as well or better than existing empirical predictors, depending on the chosen error metric. We add this new predictor into an established model for inner-shelf sorted bedforms. Additionally we incorporate a previously reported machine-learning-derived predictor for oscillatory flow ripples into the sorted bedform model. This new "hybrid" sorted bedform model, whereby machine learning components are integrated into a numerical model, demonstrates a method of incorporating observational data (filtered through a machine learning algorithm) directly into a numerical model. Results suggest that the new hybrid model is able to capture dynamics previously absent from the model -specifically, two observed pattern modes of sorted bedforms. Lastly we discuss the challenge of integrating data-driven components into morphodynamic models and the future of hybrid modeling.

show abstract

“…e interest in studying generalization and over ing in GP has been recently increasing [1,3,4,[6][7][8][9]. Geometric Semantic Genetic Programming (GSGP) [13] has also contributed to this rising interest by de ning a set of variation operators that have been shown to perform more e ectively than the corresponding Standard GP operators [4,13].…”

Section: Introductionmentioning

confidence: 99%

Unsure when to stop?

Gonçalves

Silva

Fonseca

et al. 2017

Proceedings of the Genetic and Evolutionary Computation Conference

Self Cite

View full text Add to dashboard Cite

In iterative supervised learning algorithms it is common to reach a point in the search where no further induction seems to be possible with the available data. If the search is continued beyond this point, the risk of over ing increases signi cantly. Following the recent developments in inductive semantic stochastic methods, this paper studies the feasibility of using information gathered from the semantic neighborhood to decide when to stop the search. Two semantic stopping criteria are proposed and experimentally assessed in Geometric Semantic Genetic Programming (GSGP) and in the Semantic Learning Machine (SLM) algorithm (the equivalent algorithm for neural networks). e experiments are performed on real-world high-dimensional regression datasets.e results show that the proposed semantic stopping criteria are able to detect stopping points that result in a competitive generalization for both GSGP and SLM. is approach also yields computationally e cient algorithms as it allows the evolution of neural networks in less than 3 seconds on average, and of GP trees in at most 10 seconds. e usage of the proposed semantic stopping criteria in conjunction with the computation of optimal mutation/learning steps also results in small trees and neural networks.

show abstract

Balancing Learning and Overfitting in Genetic Programming with Interleaved Sampling of Training Data

Cited by 60 publications

References 13 publications

Unveiling the properties of structured grammatical evolution

Unveiling the properties of structured grammatical evolution

Data-driven components in a model of inner-shelf sorted bedforms: a new hybrid model

Unsure when to stop?

Contact Info

Product

Resources

About