arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2404.06120physics.app-phcond-mat.mtrl-sci
收藏

通过从过往文献中学习预测任意化学计量无机材料的未来应用

Predicting the future applications of any stoichiometric inorganic material through learning from past literature

Yu Wu, Teng Liu, Haiyang Song, Yinghe Zhao, Jinxing Gu, Kailang Liu, Huiqiao Li, Jinlan Wang, Tianyou Zhai

AI总结:

针对既有模型只能预测文献中已知材料应用的问题,本文构建可生成任意化学计量无机材料词嵌入的模型,实现仅用笔记本电脑预测新材料未来应用,并经历史验证具有高可靠性。

摘要:

通过从过往文献中学习,人工智能模型已经能够预测各种化学计量无机材料在材料科学多个子领域中的未来应用。这种能力为加速新型功能材料的研发(R&D)提供了令人振奋的机遇。遗憾的是,以往模型只能对过往文献中已有材料给出预测,却无法预测试种材料的应用。本文构建了一个能够预测任意化学计量无机材料应用的模型,无论其是否为新材料。历史验证证实了该模型具有很高的可靠性。该模型的关键在于,它能够生成任意化学计量无机材料的词嵌入,而这是以往模型无法实现的。本工作构建了一个强大的模型,仅使用一台笔记本电脑即可预测任意化学计量无机材料的未来应用,有望革新新型功能材料的研发范式。

英文摘要:

Through learning from past literature, artificial intelligence models have been able to predict the future applications of various stoichiometric inorganic materials in a variety of subfields of materials science. This capacity offers exciting opportunities for boosting the research and development (R&D) of new functional materials. Unfortunately, the previous models can only provide the prediction for existing materials in past literature, but cannot predict the applications of new materials. Here, we construct a model that can predict the applications of any stoichiometric inorganic material (regardless of whether it is a new material). Historical validation confirms the high reliability of our model. Key to our model is that it allows the generation of the word embedding of any stoichiometric inorganic material, which cannot be achieved by the previous models. This work constructs a powerful model, which can predict the future applications of any stoichiometric inorganic material using only a laptop, potentially revolutionizing the R&D paradigm for new functional materials

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑