Scholarlink

<p>Additively manufactured high-entropy alloys (AMed HEAs) are promising aerospace materials, but limited experimental data hinder reliable property prediction. This study presents a transfer learning (TL) strategy in which a neural network is pre-trained on a large as-cast HEA dataset and fine-tuned using a small AMed dataset for hardness prediction. Results show that the TL model outperformed conventional machine learning methods, demonstrating its effectiveness in overcoming data scarcity.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Leveraging transfer learning to address materials data scarcity: a case study on predicting hardness in additively manufactured high-entropy alloys

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月6日

查看全文
  • Libin Zhang,
  • Dongwon Shin,
  • Dennis M. Dimiduk,
  • Yoon Suk Choi

摘要

Additively manufactured high-entropy alloys (AMed HEAs) are promising aerospace materials, but limited experimental data hinder reliable property prediction. This study presents a transfer learning (TL) strategy in which a neural network is pre-trained on a large as-cast HEA dataset and fine-tuned using a small AMed dataset for hardness prediction. Results show that the TL model outperformed conventional machine learning methods, demonstrating its effectiveness in overcoming data scarcity.