MODis
收藏资源简介:
MODis是由凯斯西储大学研究团队开发的一种多目标数据发现框架,旨在生成能够在多个用户定义的性能指标上优化模型性能的数据集。该框架将数据集成和机器学习模型性能估计相结合,以追求多目标数据发现的范式。MODis能够针对给定的数据源和模型,生成满足多个性能指标的 skyline 数据集。论文中介绍了MODis的正式计算模型,以及三种可行算法来生成 skyline 数据集,并通过实验验证了算法的有效性。
MODis is a multi-objective data discovery framework developed by the research team at Case Western Reserve University, which aims to generate datasets that optimize model performance across multiple user-defined performance metrics. This framework combines data integration and machine learning model performance estimation to pursue the paradigm of multi-objective data discovery. MODis can generate skyline datasets that meet multiple performance metrics for given data sources and models. The paper introduces the formal computational model of MODis, three feasible algorithms for generating skyline datasets, and verifies the effectiveness of the algorithms via experiments.

- 1Generating Skyline Datasets for Data Science Models凯斯西储大学 · 2025年



