Evolutionary Multitask Optimization: a Methodological Overview, Challenges, and Future Research Directions

Osaba, Eneko; Ser, Javier Del; Martinez, Aritz D.; Hussain, Amir

doi:10.1007/s12559-022-10012-8

Cited by 47 publications

(10 citation statements)

References 131 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…After years of activity that have been summarized in recent surveys on Evolutionary Multitasking [7,8], we firmly believe that it is the moment to expose and reflect these crucial concerns. Solid and informed answers to these fundamental questions are still lacking, which can lead to undesirable developments and outcomes of no practical value in the future of this field.…”

Section: How?mentioning

confidence: 99%

“…For further information about how algorithmic schemes based on these two strategies work, we refer our readers to comprehensive surveys recently published in [7,8,11,12]. Among them, MFEA [10] and MFEA-II [13] stand out as the arguably most influential works in the field.…”

Section: Evolutionary Multitask Optimization: Concepts and Relationsh...mentioning

confidence: 99%

“…An upsurge of contributions has been noted around Evolutionary Multitasking and related concepts, as has been made clear in recent survey papers around the topic [7,8]. However, some fundamental questions need still to be clarified.…”

Section: Fundamental Issues In Evolutionary Multitask Optimizationmentioning

confidence: 99%

See 2 more Smart Citations

Evolutionary Multitask Optimization: Fundamental research questions, practices, and directions for the future

Osaba¹,

Ser²,

Suganthan³

2022

Swarm and Evolutionary Computation

Self Cite

View full text Add to dashboard Cite

Section: How?mentioning

confidence: 99%

Section: Evolutionary Multitask Optimization: Concepts and Relationsh...mentioning

confidence: 99%

See 1 more Smart Citation

Evolutionary Multitask Optimization: Fundamental research questions, practices, and directions for the future

Osaba¹,

Ser²,

Suganthan³

2022

Swarm and Evolutionary Computation

Self Cite

View full text Add to dashboard Cite

“…Cheng et al [27] showed that learning a global representation of the features from multiple tasks is more efficient than learning feature representations for each task individually. Other investigators have considered the use of evolutionary multitask optimization [28] techniques to solve different tasks simultaneously. In contrast to multitask learning, evolutionary multitask optimization does not involve learning a global representation of features from the tasks.…”

Section: Introductionmentioning

confidence: 99%

“…Rather, knowledge transfer between the tasks occurs in the space of decision variables through operations that are usually heuristically defined. As noted in [28], the efficiency of any procedure in solving multiple optimization tasks largely depends on its ability to learn and exploit useful global information among tasks. Yao et al [29] applied a multifactorial evolutionary algorithm for production optimization involving reservoirs with different rock and fluid properties, but similar well patterns.…”

Section: Introductionmentioning

confidence: 99%

Multi-Asset Closed-Loop Reservoir Management Using Deep Reinforcement Learning

Nasir¹,

Durlofsky²

2022

Preprint

View full text Add to dashboard Cite

Closed-loop reservoir management (CLRM), in which history matching and production optimization are performed multiple times over the life of an asset, can provide significant improvement in the specified objective. These procedures are computationally expensive due to the large number of flow simulations required for data assimilation and optimization. Existing CLRM procedures are applied asset by asset, without utilizing information that could be useful over a range assets. Here, we develop a CLRM framework for multiple assets with varying numbers of wells. We use deep reinforcement learning to train a single global control policy that is applicable for all assets considered. The new framework is an extension of a recently introduced control policy methodology for individual assets. Embedding layers are incorporated into the representation to handle the different numbers of decision variables that arise for the different assets. Because the global control policy learns a unified representation of useful features from multiple assets, it is less expensive to construct than asset-by-asset training (we observe about 3× speedup in our examples). The production optimization problem includes a relative-change constraint on the well settings, which renders the results suitable for practical use. We apply the multi-asset CLRM framework to 2D and 3D water-flooding examples. In both cases, four assets with different well counts, well configurations, and geostatistical descriptions are considered. Numerical experiments demonstrate that the global control policy provides objective function values, for both the 2D and 3D

show abstract