...
  • 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Journal of Non-Crystalline Solids: A Journal Devoted to Oxide, Halide, Chalcogenide and Metallic Glasses, Amorphous Semiconductors, Non-Crystalline Films, Glass-Ceramics and Glassy Composites >Machine learning on density and elastic property of oxide glasses driven by large dataset
【24h】

Machine learning on density and elastic property of oxide glasses driven by large dataset

机译：大型数据集驱动氧化玻片密度和弹性性能的机器学习

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

We conduct a comprehensive machine learning study on predicting the oxide glasses density, Young's modulus, shear modulus, and Poisson's ratio given compositions, by leveraging the large dataset collected at Corning Incorporated. The density and elastic property datasets have the size of 24,858 and 8519, respectively. Our results show that random forest, K-Nearest neighbor, and neural networks consistently deliver good performance, while support vector machine consistently underperforms. Lasso linear regression works best for density prediction, while Poisson's ratio is extremely challenging to predict with the best R-2 score achieved to be around 0.7 by random forest. Additionally, feature importance analysis shows that La2O3 and BaO are the top two features for density prediction, while Na2O and B2O3 for Young's modulus and shear modulus prediction, and CaO and SiO2 for Poisson's ratio prediction. The study could be potentially leveraged as the baseline by future studies given the large dataset we trained on.
机译：通过利用在康宁掺入的大型数据集中，我们开展了预测氧化玻璃密度，杨氏模量，剪切模量和泊松比的综合机器学习研究。密度和弹性属性数据集分别具有24,858和8519的大小。我们的研究结果表明，随机森林，K最近邻居和神经网络一贯提供良好的性能，而支持向量机始终如一的表现。套索线性回归最适合密度预测，而泊松的比率非常具有挑战性，以预测随机森林约为0.7的最佳R-2得分。此外，特征重要性分析表明，LA2O3和BAO是密度预测的前两个特征，而NA2O和B2O3用于杨氏模量和剪切模量预测，以及用于泊松比预测的CAO和SIO2。由于我们培训的大型数据集，可以将该研究潜在地利用作为基线。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号