A Novel Neural Network-Based Symbolic Regression Method: Neuro-Encoded Expression Programming

Anjum, Aftab; Sun, Fengyang; Wang, Lin; Orchard, Jeff

doi:10.1007/978-3-030-30484-3_31

Cited by 8 publications

(4 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Performance on Real-world Problems We validate the performance of OPT-GAN by two real-world problems. One of them is the optimization of neural network-based symbolic regresser: NEEP (Anjum et al 2019), which is a high dimensional optimization problem with very complex landscape. Another is Frequency Modulated Sounds Parameter Identification (FMSPI) problem (Herrera and Lozano 2000), which plays an essential role in modern music, e.g., emulation of acoustic musical instruments.…”

Section: Comparison With Traditional Optimizersmentioning

confidence: 99%

OPT-GAN: A Broad-Spectrum Global Optimizer for Black-Box Problems by Learning Distribution

Ning

Liu

et al. 2023

AAAI

View full text Add to dashboard Cite

Black-box optimization (BBO) algorithms are concerned with finding the best solutions for problems with missing analytical details. Most classical methods for such problems are based on strong and fixed a priori assumptions, such as Gaussianity. However, the complex real-world problems, especially when the global optimum is desired, could be very far from the a priori assumptions because of their diversities, causing unexpected obstacles. In this study, we propose a generative adversarial net-based broad-spectrum global optimizer (OPT-GAN) which estimates the distribution of optimum gradually, with strategies to balance exploration-exploitation trade-off. It has potential to better adapt to the regularity and structure of diversified landscapes than other methods with fixed prior, e.g., Gaussian assumption or separability. Experiments on diverse BBO benchmarks and high dimensional real world applications exhibit that OPT-GAN outperforms other traditional and neural net-based BBO algorithms. The code and Appendix are available at https://github.com/NBICLAB/OPT-GAN

show abstract

Section: Comparison With Traditional Optimizersmentioning

confidence: 99%

OPT-GAN: A Broad-Spectrum Global Optimizer for Black-Box Problems by Learning Distribution

Ning

Liu

et al. 2023

AAAI

View full text Add to dashboard Cite

show abstract

“…Anjum, Sun, Wang, and Orchard [28] propose novel programming constructs to enable the integration of EMIA, an advanced emotion model based on rules, with the 2APL agent language. Through a reconsideration of 2APL's syntax, semantics, and decision-making process, we have successfully integrated them.…”

Section: Agent Programming Languages Extensionsmentioning

confidence: 99%

An Analysis of Multi Agent Systems Agent Based Programming

Кhusein

2023

JCNS

View full text Add to dashboard Cite

The effectiveness of agent-based modeling as a simulation modeling methodology has resulted in its application in diverse settings, including the resolution of pragmatic business challenges, in recent times. The domain of symbolic artificial intelligence, which investigates intelligent and self-governing entities, is preoccupied with the mechanisms by which these entities arrive at determinations regarding their conduct in reaction to, or in expectation of, stimuli from the external environment. The scope of the methods employed encompasses a diverse array of techniques, spanning from negotiations to agent simulations, as well as multi-agent argumentation and planning. The present article scrutinizes the utilization of agent-based computing in multi-agent systems and provides an all-encompassing analysis of the relevant literature. This study delves into the examination of both traditional and contemporary agent programming languages, including their respective extensions, comparative analyses, and instances of their application in published literature.

show abstract

“…These methods propose to take advantage of the computational capacity of neural networks (Hornik et al, 1989) while providing an interpretable solution. For instance, they offer to encode the expression in the neural network structure and activation functions (Sahoo et al, 2018;Kim et al, 2020), to predict a string expression (Anjum et al, 2019) with Recurrent Neural Networks, or to use Deep Reinforcement Learning (Deep RL) as a search engine (Petersen et al, 2021). By combining partial derivative and neural networks, AI Feynman method (Udrescu and Tegmark, 2020) propose to make use of simplifying properties (such as units, symmetry, separability..…”

Section: Symbolic Regressionmentioning

confidence: 99%

A Reinforcement Learning Approach to Domain-Knowledge Inclusion Using Grammar Guided Symbolic Regression

Crochepierre¹,

Boudjeloud-Assala²,

Barbesant³

2022

Preprint

View full text Add to dashboard Cite

In recent years, symbolic regression has been of wide interest to provide an interpretable symbolic representation of potentially large data relationships. Initially circled to genetic algorithms, symbolic regression methods now include a variety of Deep Learning based alternatives. However, these methods still do not generalize well to real-world data, mainly because they hardly include domain knowledge nor consider physical relationships between variables such as known equations and units. Regarding these issues, we propose a Reinforcement-Based Grammar-Guided Symbolic Regression (RBG2-SR) method that constrains the representational space with domain-knowledge using context-free grammar as reinforcement action space. We detail a Partially-Observable Markov Decision Process (POMDP) modeling of the problem and benchmark our approach against state-of-the-art methods. We also analyze the POMDP state definition and propose a physical equation search use case on which we compare our approach to grammar-based and non-grammarbased symbolic regression methods. The experiment results show that our method is competitive against other state-of-the-art methods on the benchmarks and offers the best error-complexity trade-off, highlighting the interest of using a grammar-based method in a real-world scenario.

show abstract

A Novel Neural Network-Based Symbolic Regression Method: Neuro-Encoded Expression Programming

Cited by 8 publications

References 34 publications

OPT-GAN: A Broad-Spectrum Global Optimizer for Black-Box Problems by Learning Distribution

OPT-GAN: A Broad-Spectrum Global Optimizer for Black-Box Problems by Learning Distribution

An Analysis of Multi Agent Systems Agent Based Programming

A Reinforcement Learning Approach to Domain-Knowledge Inclusion Using Grammar Guided Symbolic Regression

Contact Info

Product

Resources

About