Dynamic multiobjective optimization driven by inverse reinforcement learning

Zou, Fei; Yen, Gary G.; Zhao, Chen

doi:10.1016/j.ins.2021.06.054

Cited by 22 publications

(6 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition to the three types of methods mentioned above, there are some extraordinary methods based on machine learning and reinforcement learning mechanisms [52][53][54]. Zou et al [55] proposed a dynamic multi-objective evolutionary algorithm driven by inverse reinforcement learning. This algorithm uses limited data to accomplish a promising evolutionary process and to guide the search for populations.…”

Section: Diversity-based Methodsmentioning

confidence: 99%

Combining Key-Points-Based Transfer Learning and Hybrid Prediction Strategies for Dynamic Multi-Objective Optimization

Wang

2022

Mathematics

View full text Add to dashboard Cite

Dynamic multi-objective optimization problems (DMOPs) have been of interest to many researchers. These are problems in which the environment changes during the evolutionary process, such as the Pareto-optimal set (POS) or the Pareto-optimal front (POF). This kind of problem imposes more challenges and difficulties for evolutionary algorithms, mainly because it demands population to track the changing POF efficiently and accurately. In this paper, we propose a new approach combining key-points-based transfer learning and hybrid prediction strategies (KPTHP). In particular, the transfer process combines predictive strategy with obtaining anticipated key points depending on the previous moments to acquire the optimal individuals at the new instance during the evolution. Additionally, center-point-based prediction is used to complement transfer learning to comprehensively generate initial populations. KPTHP and six state-of-the-art algorithms are tested on various test functions for MIGD, DMIGD, MMS, and HVD metrics. KPTHP obtains superior results on most of the tested functions, which shows that our algorithm performs excellently in both convergence and diversity, with more competitiveness in addressing dynamic problems.

show abstract

Section: Diversity-based Methodsmentioning

confidence: 99%

Combining Key-Points-Based Transfer Learning and Hybrid Prediction Strategies for Dynamic Multi-Objective Optimization

Wang

2022

Mathematics

View full text Add to dashboard Cite

show abstract

“…This method detects the type of change and then selects an appropriate prediction model to generate an initial population for the new environment. Indeed, these prediction-based DMOEAs can estimate the change trends in different environments and have shown promising performance in solving various kinds of DMOPs [16,50]. However, most of them assume that the data used to construct prediction models obey the IID condition, which may not always hold true in some practical cases and will affect the prediction accuracy.…”

Section: Related Workmentioning

confidence: 99%

“…The prediction-based Multiple source transfer learning for dynamic multiobjective optimization DMOEAs will learn prediction models from historical environments, which can handle various dynamic changes in DMOPs by predicting their change tendency. They have been validated to perform better than other types of DMOEAs for solving MOPs with more kinds of dynamic changes [43,50]. However, in most prediction-based DMOEAs, the historical solutions used to construct prediction models are assumed to obey an independent identical distribution (IID), which may not always hold true in some practical cases.…”

Section: Introductionmentioning

confidence: 99%

Multiple source transfer learning for dynamic multiobjective optimization

Lin

et al. 2022

Information Sciences

View full text Add to dashboard Cite

“…We consider two measures of experimental comprehensiveness: the number of combinations and the range of each parameter. The number of examined 𝑛 𝑡 -𝜏 𝑡 pairings is varied in recent works including one [20,32], two [23,36], three [13,22,38,43] and five [21]. Only the recent work by Zhang et al [37] considers a more diverse set with six different frequency-severity pairings: 𝑛 𝑡 -𝜏 𝑡 = {5 − 5, 5 − 10, 5 − 20, 10 − 5, 10 − 10, 10 − 20}.…”

Section: Severity and Frequencymentioning

confidence: 99%

“…The NSGA-III algorithm is also used and modified for performance comparisons [37], as is the SPEA2 [42] algorithm [21,22,25]. The non-dynamic MOEA/D algorithm [39] is also compared [8,20,22] and augmented with Kalman Filter prediction [8,13], reinforcement learning [43], intensity of environmental change handling [32,43] & a first order difference model [37].…”

Section: Use Of Non-dynamic Moeas For Dmopsmentioning

confidence: 99%

Reproducibility and Baseline Reporting for Dynamic Multi-objective Benchmark Problems

Herring¹,

Kirley²,

Yao³

2022

Preprint

View full text Add to dashboard Cite

Dynamic multi-objective optimization problems (DMOPs) are widely accepted to be more challenging than stationary problems due to the time-dependent nature of the objective functions and/or constraints. Evaluation of purpose-built algorithms for DMOPs is often performed on narrow selections of dynamic instances with differing change magnitude and frequency or a limited selection of problems. In this paper, we focus on the reproducibility of simulation experiments for parameters of DMOPs. Our framework is based on an extension of PlatEMO, allowing for the reproduction of results and performance measurements across a range of dynamic settings and problems. A baseline schema for dynamic algorithm evaluation is introduced, which provides a mechanism to interrogate performance and optimization behaviours of well-known evolutionary algorithms that were not designed specifically for DMOPs. Importantly, by determining the maximum capability of non-dynamic multi-objective evolutionary algorithms, we can establish the minimum capability required of purpose-built dynamic algorithms to be useful. The simplest modifications to manage dynamic changes introduce diversity. Allowing non-dynamic algorithms to incorporate mutated/random solutions after change events determines the improvement possible with minor algorithm modifications. Future expansion to include current dynamic algorithms will enable reproduction of their results and verification of their abilities and performance across DMOP benchmark space.

show abstract

Dynamic multiobjective optimization driven by inverse reinforcement learning

Cited by 22 publications

References 34 publications

Combining Key-Points-Based Transfer Learning and Hybrid Prediction Strategies for Dynamic Multi-Objective Optimization

Combining Key-Points-Based Transfer Learning and Hybrid Prediction Strategies for Dynamic Multi-Objective Optimization

Multiple source transfer learning for dynamic multiobjective optimization

Reproducibility and Baseline Reporting for Dynamic Multi-objective Benchmark Problems

Contact Info

Product

Resources

About