MoleculeNet
收藏资源简介:
MoleculeNet是由斯坦福大学的研究团队开发的一个大规模分子机器学习基准。该数据集整合了多个公共数据源,涵盖了超过70万个化合物的多种属性,包括量子力学、物理化学、生物物理和生理效应等四个层次的分子特性。MoleculeNet不仅提供了数据集,还提供了评估指标、高质量的开源实现算法以及数据分割方法,旨在为分子机器学习领域提供一个标准化的评估平台。通过这个平台,研究者可以更容易地开发和改进用于学习分子属性的模型,从而推动化学和机器学习领域的进步。
MoleculeNet is a large-scale molecular machine learning benchmark developed by a research team at Stanford University. This dataset integrates multiple public data sources, covering diverse properties of over 700,000 compounds, including four hierarchical molecular characteristics: quantum mechanics, physical chemistry, biophysics, and physiological effects. In addition to providing the dataset itself, MoleculeNet also offers evaluation metrics, high-quality open-source implementation algorithms, and data splitting methods, aiming to provide a standardized evaluation platform for the field of molecular machine learning. Through this platform, researchers can more easily develop and refine models for learning molecular properties, thereby advancing the progress of the fields of chemistry and machine learning.




