Prediction of PbTe-like Thermoelectric Materials’ Properties with a Neural Network

Riku Sato; Takushi Kodani; Yukari Katsura; Masaya Kumagai; Yoji Imai; Sakiko Gunji; Kaoru Kimura

Presentation information

Oral presentation

9 Applied Materials Science » 9.4 Thermoelectric conversion

[17p-F102-1~17] 9.4 Thermoelectric conversion

Sat. Mar 17, 2018 1:45 PM - 6:30 PM F102 (61-102)

Kaoru Toko(Univ. of Tsukuba), Yoshiaki Nakamura(Osaka Univ.), Kei Hayashi(Tohoku Univ.), Yukari Katsura(Univ. of Tokyo)

6:00 PM - 6:15 PM

△ [17p-F102-16] Prediction of PbTe-like Thermoelectric Materials’ Properties with a Neural Network

〇Riku Sato¹, Takushi Kodani^1,2, Yukari Katsura^1,2, Masaya Kumagai³, Yoji Imai^2,3, Sakiko Gunji², Kaoru Kimura¹ (1.Tokyo Univ., 2.NIMS, 3.RIKEN)

Keywords:thermoelectric materials, machine learning, neural network

By using experimental data on PbTe-like thermoelectric materials on 66 papers, machine learning and properties prediction were performed. Our neural network succeeded in prediction of thermoelectric properties, such as Seebeck coefficient and thermal conductivity from composition within the mean error range of 10%.