CFP: Integrating Numerical Optimization Methods with Genetic Programming

[Full CFP on the Journal's Springer site]

Guest Editors

• Anna I Esparcia-Alcázar
  Universitat Politècnica de València, Spain: aesparcia@ieee.org
• Leonardo Trujillo
  Instituto Tecnológico de Tijuana, Mexico: leonardo.trujillo@tectijuana.edu.mx

About this Issue

This special issue focuses on integrating numerical optimization methods with Genetic Programming (GP) in order to improve the evolutionary search. In traditional GP the search space is the space of all possible syntactic expressions that can be generated from the set of functions and terminals, which depends upon the type of program representation used. The search operators modify individuals at the level of syntax, and, given that syntactic expressions tend to be fragile, their effect on behavior is usually non-local and difficult to predict. This has lead researchers to explore other search operators or program representations.

One possibility is to use numerical optimization methods as a local search process. In fact, the representation and adaptation (i.e. learning) of real-valued parameters in GP is still an open issue in GP at large, where most of the work has focused on what Koza termed ephemeral random constants and some effort has been devoted to adapting them [1]. Although more advanced approaches such as adaptive node gains were proposed over two decades ago [2] and their uptake has produced some success [3] [4], it has been relatively limited [5]. This area is particularly relevant in modern machine learning, where powerful computing platforms like GPUs are highly optimized for performing such tasks. Numerical optimization can also be used to tune hyper-parameters or to derive surrogate models on-line.

This special issue intends to explore new representations, algorithms and methodologies that can enhance GP systems by exploiting numerical optimization techniques to improve convergence, reduce computation cost and achieve state-of-the- art performance in real-world machine learning challenges [6], and in particular in Deep Learning, a field in which Genetic Programming is becoming increasingly successful [7].

Scope

• Novel representations that are amenable to numerical and local search methods
• Search approaches for real-valued parameters or meta-parameters in GP individuals
• New search operators that can exploit both syntactic and numerical search
• Implementations that improve search efficiency and reduce training times
• Techniques that are optimized for High Performance Computing platforms, such as GPUs and FPGAs

Important Dates

• Submission deadline: January 20, 2019 
• Notification of first review: May 2, 2019 
• Resubmission: June 3, 2019
• Final acceptance notification: August 2, 2019

Submissions and Review Procedures

Special Issues are handled in the normal way via the online Editorial Manager system found at https://genp.edmgr.com. Please choose the article type “Integrating Numerical Optimization Methods with Genetic Programming.” Special Issue articles should fulfil all the standard requirements of any GPEM article. Authors should note that the same criteria apply to articles in Special Issues as to regular articles. Special Issue articles must not consist of overviews of the authors' previously published work, e.g. peer- reviewed articles, book chapters, official reports, etc.

All papers will undergo the same rigorous GPEM review process. Please refer to the GPEM website for detailed instructions on paper submission: http://www.springer.com/10710

References

[1] L. M. Howard and D. J. D'Angelo, "The GA-P: a genetic algorithm and genetic programming hybrid," in IEEE Expert, vol. 10, no. 3, pp. 11-15, Jun 1995. doi: http://dx.doi.org/10.1109/64.393137

[2] Esparcia-Alcázar A.I., Sharman K.C. (1996) Genetic programming techniques that evolve recurrent neural network architectures for signal processing, Neural Networks for Signal Processing VI. Proceedings of the 1996 IEEE Signal Processing Society Workshop, pages 139-148 DOI: http://dx.doi.org/10.1109/NNSP.1996.548344

[3] Maarten Keijzer. 2004. Scaled Symbolic Regression. Genetic Programming and Evolvable Machines 5, 3 (September 2004), 259-269. DOI: http://dx.doi.org/10.1023/B:GENP.0000030195.77571.f9

[4] Emigdio Z-Flores, Leonardo Trujillo, Oliver Schütze, and Pierrick Legrand. 2015. A Local Search Approach to Genetic Programming for Binary Classification. In Proceedings of the 2015 Annual Conference on Genetic and Evolutionary Computation (GECCO '15), Sara Silva (Ed.). ACM, New York, NY, USA, 1151-1158. DOI: http://dx.doi.org/10.1145/2739480.2754797

[5] Leonardo Trujillo, Emigdio Z-Flores, Perla S. Juárez Smith, Pierrick Legrand, Sara Silva, Mauro Castelli, Leonardo Vanneschi, Oliver Schütze and Luis Muñoz. Local Search is Underused in Genetic Programming. In Rick Riolo et al. editors, Genetic Programming Theory and Practice XIV, Ann Arbor, USA, 2017. Springer.

[6] Numerical and Evolutionary Optimization Workshop: http://neo.cinvestav.mx/NEO2018/

[7] Risto Miikkulainen, Evolving Multitask Neural Network Structure, Metalearning Symposium at NIPS 2017, http://metalearning-symposium.ml/files/miikkulainen.pdf