A novel direct measure of exploration and exploitation based on attraction basins

Jerebic, Jernej; Mernik, Marjan; Liu, Shih-Hsi; Ravber, Miha; Baketarić, Mihael; Mernik, Luka; Črepinšek, Matej

doi:10.1016/j.eswa.2020.114353

Cited by 41 publications

(23 citation statements)

References 54 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Typically, there is an inherent trade-off between these two problems, named as the learning/earning trade-off (Rothschild 1974;Cheung et al 2017). It is akin to the trade-off of exploration versus exploitation that we encounter in machine learning and evolutionary optimization (Tokic 2010;Crepinšek et al 2013;Rezaei and Safavi 2020;Jerebic et al 2021;Mahesh and Sushnigdha 2021).…”

Section: Introductionmentioning

confidence: 99%

Novel pricing strategies for revenue maximization and demand learning using an exploration–exploitation framework

2021

View full text Add to dashboard Cite

The price demand relation is a fundamental concept that models how price affects the sale of a product. It is critical to have an accurate estimate of its parameters, as it will impact the company's revenue. The learning has to be performed very efficiently using a small window of a few test points, because of the rapid changes in price demand parameters due to seasonality and fluctuations. However, there are conflicting goals when seeking the two objectives of revenue maximization and demand learning, known as the learn/earn trade-off. This is akin to the exploration/exploitation trade-off that we encounter in machine learning and optimization algorithms. In this paper, we consider the problem of price demand function estimation, taking into account its exploration-exploitation characteristic. We design a new objective function that combines both aspects. This objective function is essentially the revenue minus a term that measures the error in parameter estimates. Recursive algorithms that optimize this objective function are derived. The proposed method outperforms other existing approaches.

show abstract

Section: Introductionmentioning

confidence: 99%

Novel pricing strategies for revenue maximization and demand learning using an exploration–exploitation framework

2021

View full text Add to dashboard Cite

show abstract

“…EAs are population-based search algorithms, which mimic concepts from biological evolution, such as survival of the fittest, crossover, and mutation. EAs are known to have a remarkable balance between exploration and exploitation [22] [51], which is needed to search an enormous space of all possible solutions efficiently [29], [21] [32]. Early examples of EAs are Genetic Algorithms (GAs), Evolutionary Strategies (ES), and Genetic Programming (GP) [26].…”

Section: Introductionmentioning

confidence: 99%

A JSSP solution for production planning optimization combining industrial engineering and evolutionary algorithms

Srsen

Mernik

2021

ComSIS

Self Cite

View full text Add to dashboard Cite

A Job Shop Scheduling Problem (JSSP), where p processes and n jobs should be processed on m machines so that the total completion time is minimal, is a well-known problem with many industrial applications. Many researchers focus on the JSSP classification and algorithms that address the different JSSP classes. In this research work, the production times, a very well-known theme covered in Industrial Engineering (IE), are integrated into an Evolutionary Algorithm (EA) to present a solution for real-world manufacturing JSSP problems solving. Since a drawback of classical IE is a manual determination of the technological times, an Internet of Things (IoT) architecture is proposed as a possible solution.

show abstract

Section: Introductionmentioning

confidence: 99%

“…Figure 2 presents a landscape, heatmap, and attraction basins computed correctly with the algorithm from [22]. Attraction basins are used, for example, to characterise the fitness landscapes (see, e.g., in [23,24]), or analyse the metaheuristic (e.g., researchers in [22,25] measure exploration). A fitness landscape is an important concept, which was borrowed from biology [26] to analyse the optimisation problems and the relation of their characteristics to the performance of metaheuristics [27][28][29][30].…”

Section: Introductionmentioning

confidence: 99%

“…The same holds for the exploration and exploitation, two distinct phases of each metaheuristic that are found to be crucial to understand and use metaheuristics efficiently [37][38][39][40][41]. In [22], attraction basins were used to measure exploration and exploitation directly. For example, in Figure 3, attraction basins are denoted with different colours, solutions given by a certain metaheuristic are denoted with red circles, and the sequence of solutions is ordered by the white arrows.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Attraction Basins in Metaheuristics: A Systematic Mapping Study

2021

Self Cite

View full text Add to dashboard Cite

Context: In this study, we report on a Systematic Mapping Study (SMS) for attraction basins in the domain of metaheuristics. Objective: To identify research trends, potential issues, and proposed solutions on attraction basins in the field of metaheuristics. Research goals were inspired by the previous paper, published in 2021, where attraction basins were used to measure exploration and exploitation. Method: We conducted the SMS in the following steps: Defining research questions, conducting the search in the ISI Web of Science and Scopus databases, full-text screening, iterative forward and backward snowballing (with ongoing full-text screening), classifying, and data extraction. Results: Attraction basins within discrete domains are understood far better than those within continuous domains. Attraction basins on dynamic problems have hardly been investigated. Multi-objective problems are investigated poorly in both domains, although slightly more often within a continuous domain. There is a lack of parallel and scalable algorithms to compute attraction basins and a general framework that would unite all different definitions/implementations used for attraction basins. Conclusions: Findings regarding attraction basins in the field of metaheuristics reveal that the concept alone is poorly exploited, as well as identify open issues where researchers may improve their research.

show abstract

A novel direct measure of exploration and exploitation based on attraction basins

Cited by 41 publications

References 54 publications

Novel pricing strategies for revenue maximization and demand learning using an exploration–exploitation framework

Novel pricing strategies for revenue maximization and demand learning using an exploration–exploitation framework

A JSSP solution for production planning optimization combining industrial engineering and evolutionary algorithms

Attraction Basins in Metaheuristics: A Systematic Mapping Study

Contact Info

Product

Resources

About