surrogate constraint
Recently Published Documents


TOTAL DOCUMENTS

29
(FIVE YEARS 1)

H-INDEX

9
(FIVE YEARS 0)

2022 ◽  
Author(s):  
Leon Faure ◽  
Bastien Mollet ◽  
Wolfram Liebermeister ◽  
Jean-Loup Faulon

Metabolic networks have largely been exploited as mechanistic tools to predict the behavior of microorganisms with a defined genotype in different environments. However, flux predictions by constraint-based modeling approaches are limited in quality unless labor-intensive experiments including the measurement of media intake fluxes, are performed. Using machine learning instead of an optimization of biomass flux - on which most existing constraint-based methods are based - provides ways to improve flux and growth rate predictions. In this paper, we show how Recurrent Neural Networks can surrogate constraint-based modeling and make metabolic networks suitable for backpropagation and consequently be used as an architecture for machine learning. We refer to our hybrid - mechanistic and neural network - models as Artificial Metabolic Networks (AMN). We showcase AMN and illustrate its performance with an experimental dataset of Escherichia coli growth rates in 73 different media compositions. We reach a regression coefficient of R2=0.78 on cross-validation sets. We expect AMNs to provide easier discovery of metabolic insights and prompt new biotechnological applications.


2018 ◽  
Vol 9 (1) ◽  
pp. 78-94
Author(s):  
Jose Humberto Ablanedo-Rosas ◽  
Cesar Rego

In Combinatorial Optimization the evaluation of heuristic algorithms often requires the consideration of multiple performance metrics that are relevant for the application of interest. Traditional empirical analysis of algorithms relies on evaluating individual performance metrics where the overall assessment is conducted by subjective judgment without the support of rigorous scientific methods. The authors propose an analytical approach based on data envelopment analysis (DEA) to rank algorithms by their relative efficiency scores that result from combining multiple performance metrics. To evaluate their approach, they perform a pilot study examining the relative performance of ten surrogate constraint algorithms for different classes of the set covering problem. The analysis shows their DEA-based approach is highly effective, establishing a clear difference between the algorithms' performances at appropriate statistical significance levels, and in consequence providing useful insights into the selection of algorithms to address each class of instances. Their approach is general and can be used with all types of performance metrics and algorithms.


Sign in / Sign up

Export Citation Format

Share Document