In an era of pervasive digitalization, the growing volume and variety of data streams poses a new challenge to the efficient running of data-drivenoptimization algorithms. Targeting scalable multiobjective evolution ...
详细信息
In an era of pervasive digitalization, the growing volume and variety of data streams poses a new challenge to the efficient running of data-drivenoptimization algorithms. Targeting scalable multiobjective evolution under large-instance data, this article proposes the general idea of using subsampled small-data tasks as helpful minions (i.e., auxiliary source tasks) to quickly optimize for large datasets-via an evolutionary multitasking framework. Within this framework, a novel computational resource allocation strategy is designed to enable the effective utilization of the minions while guarding against harmful negative transfers. To this end, an intertask empirical correlation measure is defined and approximated via Bayes' rule, which is then used to allocate resources online in proportion to the inferred degree of source-target correlation. In the experiments, the performance of the proposed algorithm is verified on: 1) sample average approximations of benchmark multiobjectiveoptimization problems under uncertainty and 2) practical multiobjective hyperparameter tuning of deep neural network models. The results show that the proposed algorithm can obtain up to about 73% speedup relative to existing approaches, demonstrating its ability to efficiently tackle real-world multiobjectiveoptimization involving evaluations on large datasets.
Interactive multiobjectiveoptimization methods incorporate preferences from a human decision maker in the optimization process iteratively. This allows the decision maker to focus on a subset of solutions, learn abou...
详细信息
Interactive multiobjectiveoptimization methods incorporate preferences from a human decision maker in the optimization process iteratively. This allows the decision maker to focus on a subset of solutions, learn about the underlying trade-offs among the conflicting objective functions in the problem and adjust preferences during the solution process. Incorporating preference information allows computing only solutions that are interesting to the decision maker, decreasing computation time significantly. Thus, interactive methods have many strengths making them viable for various applications. However, there is a lack of existing software frameworks to apply and experiment with interactive methods. We fill a gap in the optimization software available and introduce DESDEO, a modular and open source Python framework for interactive multiobjectiveoptimization. DESDEO's modular structure enables implementing new interactive methods and reusing previously implemented ones and their functionalities. Both scalarization-based and evolutionary methods are supported, and DESDEO allows hybridizing interactive methods of both types in novel ways and enables even switching the method during the solution process. Moreover, DESDEO also supports defining multiobjectiveoptimization problems of different kinds, such as data-driven or simulation-based problems. We discuss DESDEO's modular structure in detail and demonstrate its capabilities in four carefully chosen use cases aimed at helping readers unfamiliar with DESDEO get started using it. We also give an example on how DESDEO can be extended with a graphical user interface. Overall, DESDEO offers a much-needed toolbox for researchers and practitioners to efficiently develop and apply interactive methods in new ways - both in academia and industry.
暂无评论