结合小组贡献方法和半监督学习,构建机器学习模型,用于预测水污染物的基激素速率常数
Zhao Liu1, Lanyu Shang2, Kuan Huang1
1Department of Civil and Environmental Engineering, Case Western Reserve University, Cleveland, Ohio 44106, United States.
Environmental science & technology
|December 26, 2024
概括
通过扩展数据,改善了基 (HO•) 反应的机器学习模型. 新的方法显著增加了适用性领域,使超过56万种化学品的可靠预测成为可能.
科学领域:
- 环境化学环境化学
- 计算化学计算化学
- 化学动力学 化学动力学
背景情况:
- 机器学习 (ML) 模型有效地预测了含基 (HO•) 的有机化合物反应速率常数.
- 由于实验数据稀缺 (<1400条记录),以前的模型面临着有限的适用性领域 (AD).
研究的目的:
- 为解决HO•反应速率常数现有ML模型有限的AD问题.
- 通过策划更大的数据集和采用数据增强策略来提高ML模型性能并扩展AD.
主要方法:
- 策划了一个扩展的实验数据集 (初级数据集) 包含2358个动态记录.
- 利用群体贡献方法 (GCM) 和半监督学习 (SSL) 来增加数据集.
- 将147,168个新的数据点纳入最终的ML模型.
主要成果:
- 在最初的AD之外,GCM增强了化学品的模型性能.
- SSL有效地扩展了该模型的 AD.
- 最终的模型在测试组中实现了R2 = 0.77,RMSE = 0.32和MAE = 0.24.
- AD被扩大了117%,使得可靠的预测超过56万种化学物质.
结论:
- 结合的GCM和SSL方法有效地扩大了数据集,改善了ML模型的性能,并扩大了AD.
- 开发的ML模型证明了对一个庞大的化学空间的可靠预测能力.
- 这项研究提供了一个强大的方法来增强化学动力学中的ML模型,并提供了一个广泛可访问的在线预测器.
相关概念视频
01:11Radical Reactivity: Overview
2.0K
Radicals, the highly reactive species, gain stability by undergoing three different reactions. The first reaction involves a radical-radical coupling, in which a radical combines with another radical, forming a spin‐paired molecule. The second reaction is between a radical and a spin‐paired molecule, generating a new radical and a new spin‐paired molecule. The third reaction is radical decomposition in a unimolecular reaction, forming a new radical and a spin‐paired...
2.0K
01:02Radical Reactivity: Electrophilic Radicals
1.8K
Radicals adjacent to electron‐withdrawing groups are called electrophilic radicals. These radicals readily react with nucleophilic alkenes. For example, the malonate radical, in which the radical center is flanked by two electron‐withdrawing groups, reacts readily with butyl vinyl ether, which consists of an electron‐donating oxygen substituent. The reaction between electrophilic malonate radical and nucleophilic vinyl ether is favored because the radical has a...
1.8K
01:16Radical Reactivity: Nucleophilic Radicals
2.0K
Radicals adjacent to electron-donating groups are called nucleophilic radicals. These radicals readily react with electrophilic alkenes. The SOMO–LUMO interactions are the driving force for the reaction, where the high-energy SOMO of the electron-rich, nucleophilic radicals interacts with the low-energy LUMO of the electron-deficient, electrophilic alkenes. Such SOMO–LUMO interactions are the basis of reactive radical traps, affecting the selectivity in radical reactions. For...
2.0K
02:30Water: A Bronsted-Lowry Acid and Base
50.0K
The reaction between a Brønsted-Lowry acid and water is called acid ionization. For example, when hydrogen fluoride dissolves in water and ionizes, protons are transferred from hydrogen fluoride molecules to water molecules, yielding hydronium ions and fluoride ions:
50.0K
01:17Precipitation and Co-precipitation
1.7K
Precipitation and coprecipitation methods can be used to separate a mixture of ions in a solution. In qualitative inorganic analysis, ions that form sparingly soluble precipitates with the same reagent are separated based on the differences in solubility products. For example, consider the separation of Cu(II) and Fe(II) ions by precipitation as insoluble sulfides. First, copper(II) sulfide is precipitated by the addition of acidic H2S, where the dissociation of H2S is suppressed. Adding H2S...
1.7K
02:34Regioselectivity and Stereochemistry of Acid-Catalyzed Hydration
8.3K
The rate of acid-catalyzed hydration of alkenes depends on the alkene's structure, as the presence of alkyl substituents at the double bond can significantly influence the rate.
8.3K