摘要
一位新闻记者-机器人与机器学习的工作人员新闻编辑每日新闻-机器学习的研究结果在一份新的报告中讨论。根据NewsRx编辑在Oak Ridge,Ten Nessee的新闻报道，研究表明：“离子液体(ILs)是一种新型的绿色溶剂，具有广泛的应用前景，包括碳捕获和木质纤维素生物质分解。然而，由于离子液体在室温下具有高粘度，因此在工业规模上的应用仍然具有挑战性。”本研究的资金支持者包括美国能源部(DOE)、美国能源部(DOE)、美国能源部(DOE)。我们的新闻记者引用了橡树岭国家实验室的一篇研究文章:“为了开发低粘度液体系,需要系统地研究它们的定量结构-性质关系(QSPR)。在这里,我们建立了四个机器学习(ML)模型来预测在不同温度和压力范围内的粘度。”ML方法包括双因子多项式回归离子(双因子PR)、支持向量回归(SVR)、前馈神经网络KS(FFNN)、和分类boosting(CATBoost)是基于在以前的ML研究中被证明有用的特征开发的：cosmo-rs(真实溶剂的类导体筛选模型)-衍生的表面筛选电荷密度(sigma pr ofiles)。FFNN和CATBoost在预测IL粘度方面最准确，平均绝对相对偏差较低，R 2值较高。通过计算Tanimoto相似性分数来表征化学spa ce此外,采用SHapley附加解释(SHAP)分析对ML结果进行了解释,温度、离子的极性区域和离子的非极性区域是影响粘度预测的关键特征。
Abstract
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News Daily News-Research findings on Machine Learning are discussed in a new report. According to news reporting out of Oak Ridge, Ten nessee, by NewsRx editors, research stated, "Ionic liquids (ILs) are a novel gro up of green solvents with great promise for various industrial applications, inc luding carbon capture and lignocellulosic biomass deconstruction. However, the u se of ILs at the industrial scale remains challenging due to their high viscosit ies at ambient temperatures." Financial supporters for this research include United States Department of Energ y (DOE), United States Department of Energy (DOE), United States Department of E nergy (DOE). Our news journalists obtained a quote from the research from Oak Ridge National Laboratory, "To develop ILs with lower viscosities, a systematic study of their quantitative structure-property relationship (QSPR) is desirable. Here, we devel oped four machine learning (ML) models to predict viscosity at various temperatu re and pressure ranges, trained over a wide range of ILs consisting of various c ationic and anionic families. ML methods including two-factor polynomial regress ion (two-factor PR), support vector regression (SVR), feed-forward neural networ ks (FFNN), and categorical boosting (CATBoost) were developed based on features that have proven useful in previous ML studies: COSMO-RS (conductor-like screeni ng model for real solvents)-derived surface screening charge densities (sigma pr ofiles). FFNN and CATBoost were the most accurate in predicting IL viscosities w ith lower average absolute relative deviation and higher R 2 values on the test set. Tanimoto similarity scores were calculated to characterize the chemical spa ce and structural similarity of the investigated ions. Furthermore, SHapley Addi tive exPlanation (SHAP) analysis was employed to interpret the ML results. Tempe rature, the polar area of ILs, and the nonpolar regions of ions are key features that influence the viscosity predictions."
NETL
NSTL