Multi-task deep-learning optimization of trade-off properties for superior-performance Fe-based soft magnetic alloys

doi:10.1088/1674-1056/ae5b5f

Abstract Fe-based amorphous alloys are promising soft magnetic materials for developing next-generation devices with high frequency and efficiency. However, optimizing Fe-based alloys with ultra-high saturation magnetic flux density ($B_{{\rm s}}$), ultra-low coercivity ($H_{{\rm c}}$), and good glass-forming ability remains a notorious challenge owing to the vast composition space and complex trade-offs among these properties. Thus, conventional design methods face great challenges. Here, we develop a generative multi-task deep learning (GMTDL) approach to achieve simultaneous optimization of compositions and trade-off properties. The GMTDL can sufficiently exploit and share knowledge from datasets across different tasks, despite the limitations and imbalances of these datasets. Therefore, it exhibits superior performance in predicting alloys with multiple targeted properties, outperforming previous machine learning-based design strategies. Moreover, the GMTDL can also tailor compositions, providing an efficient way to regulate properties and generate desired candidates for further experimental processing. The validity and reliability of GMTDL are rigorously tested by benchmarking against Fe-based alloys reported very recently. Moreover, some new alloys with ultra-high $B_{{\rm s}}$ and ultra-low $H_{{\rm c}}$ are predicted. The optimal content windows of key elements and their synergistic effects are also unraveled, providing practical guidance. Thus, our study establishes an effective and reliable paradigm for simultaneous prediction and optimization of high-performance materials with multiple properties.

Keywords: Fe-based amorphous alloys soft magnetic materials multi-task deep learning multi-objective optimization inverse design

Received: 02 March 2026
Revised: 01 April 2026
Accepted manuscript online: 03 April 2026

PACS:	07.05.Mh	(Neural networks, fuzzy logic, artificial intelligence)
	75.50.Bb	(Fe and its alloys)
	75.50.Kj	(Amorphous and quasicrystalline magnetic materials)
	75.60.Ej	(Magnetization curves, hysteresis, Barkhausen and related effects)

Fund: Project supported by the National Natural Science Foundation of China (Grant Nos. 12574220 and 52031016).

Corresponding Authors: Mao-Zhi Li
E-mail: maozhili@ruc.edu.cn

Cite this article:

Kang-Yuan Li(李康源), Mao-Zhi Li(李茂枝), and Wei-Hua Wang(汪卫华) Multi-task deep-learning optimization of trade-off properties for superior-performance Fe-based soft magnetic alloys 2026 Chin. Phys. B 35 070705

[1] Silveyra J M, Ferrara E, Huber D L and Monson T C 2018 Science 362 eaao0195
[2] Perigo E A,Weidenfeller B, Kollar P and Fuzer J 2018 Appl. Phys. Rev. 5 031301
[3] Gutfleisch O, Willard M A, Bruck E, Chen C H, Sankar S G and Liu J P 2011 Adv. Mater. 23 821
[4] Shao L L, Bai R S, Wu Y X, Zhou J, Tong X, Peng H L, Liang T, Li X Z, Zeng Q S and Zhang B 2024 Mater. Futur. 3 025301
[5] Herzer G 2013 Acta Mater. 61 718
[6] Cullity B D and Graham C D 2008 Introduction to Magnetic Materials (New York: IEEE Press) pp. 439–476
[7] Kasai S, Namikawa M and Hiratani T 2016 JFE Tech. Rep. 21 14
[8] Snoek J L 1948 Physica 14 207
[9] Yao K F, Shi L X, Chen S Q, Shao Y, Chen N and Jia J L 2018 Acta Phys. Sin. 67 016101 (in Chinese)
[10] Shi L X, Hu X Y, Li Y H, Yuan G T and Yao K F 2021 Intermetallics 131 107116
[11] Li H X, Lu Z C, Wang S L, Wu Y and Lu Z P 2019 Prog. Mater. Sci. 103 235
[12] Yoshizawa Y, Oguma S and Yamauchi K 1988 J. Appl. Phys. 64 6044
[13] Ohta M and Hasegawa R 2017 IEEE Trans. Magn. 53 2000205
[14] Wang W H, Zhao R, Han R, et al. 2025 Mater. Futur. 4 033001
[15] Li H, Wang A, Liu T, Chen P B, He A, Li Q, Luan J H and Liu C T 2021 Mater. Today 42 49
[16] Li X S, Zhou J, Shen L Q, Sun B A, Bai H Y andWangWH 2023 Adv. Mater. 35 2205863
[17] Hart G L W, Mueller T, Toher C and Curtarolo S 2021 Nat. Rev. Mater. 6 730
[18] Schmidt J, Marques M R G, Botti S and Marques M A L 2019 npj Comput. Mater. 5 83
[19] Ye Y F, Wang Q, Lu J, Liu C T and Yang Y 2016 Mater. Today 19 349
[20] Stier S P, Kreisbeck C, Ihssen H, et al. 2024 Adv. Mater. 36 2407791
[21] Rao Z Y, Tung P Y, Xie R W, et al. 2022 Science 378 78
[22] Szymanski N J, Rendy B, Fei Y X, et al. 2023 Nature 624 86
[23] Zhou Z Q, He Q F, Liu X D, Wang Q, Luan J H, Liu C T and Yang Y 2021 npj Comput. Mater. 7 138
[24] Tang Y C, Wan Y, Wang Z Q, Zhang C, Han J N, Hu C H and Tang C Y 2022 Mater. Des. 219 110726
[25] Pang B, Long Z L, Long T, He R, Liu XWand Pang M W 2023 Mater. Des. 231 112054
[26] Lu Z C, Chen X, Liu X J, Lin D Y, Zhang Y B, Wang H, Jiang S H, Li H X, Wang X Z and Lu Z P 2020 npj Comput. Mater. 6 187
[27] Ruder S 2017 arXiv:1706.05098 [cs.LG]
[28] Caruana R A 1993 Proceedings of the Tenth International Conference on Machine Learning, June 27–29, 1993, Amherst, MA, USA, pp. 41– 48
[29] Ling F H, Luo J J, Li Y, Tang T, Bai L, Ouyang W L and Yamagata T 2022 Nat. Commun. 13 7681
[30] Wu Y D, Zhu Y, Wang Y X and Chiribella G 2024 Nat. Commun. 15 8796
[31] Tolstikhin I, Bousquet O, Gelly S and Schölkopf B 2017 arXiv:1711.01558 [stat.ML]
[32] Deb K and Jain H 2014 IEEE Trans. Evol. Comput. 18 577
[33] Gretton A, Borgwardt K, Rasch M J, Schölkopf B and Smola A J 2012 J. Mach. Learn. Res. 13 723
[34] Jazzbin J 2020 Geatpy: The Genetic and Evolutionary Algorithm Toolbox With High Performance in Python.
[35] Zeni C, Pinsler R, Zügner D, et al. 2025 Nature 639 624
[36] Belkina A C, Ciccolella C O, Anno R, Halpert R, Spidlen J and Snyder- Cappione J E 2019 Nat. Commun. 10 5415
[37] Yang A, Su Y, Wang Z H, Jin S M, Ren J Z, Zhang X P, Shen W F and Clark J H 2021 Green Chem. 23 4451
[38] Saha B, Gupta S, Phung D and Venkatesh S 2015 Knowl. Inf. Syst. 46 315
[39] Lever J, Krzywinski M and Altman N 2017 Nat. Methods 14 641
[40] Guo W H, Wu Y, Shi L X, Jia J L, Wang R B, Bu H T, Zhu Z F, Shao Y and Yao K F 2025 Acta Mater. 285 120643
[41] Ohta M and Yoshizawa Y 2011 J. Phys. D: Appl. Phys. 44 064004
[42] Yu M and Kakehashi Y 1994 Phys. Rev. B 49 15723
[43] Yang S Y, Zang B W, Xiang M L, Shen F Y, Song L J, Gao M, Zhang Y, Huo J T and Wang J Q 2025 Adv. Funct. Mater. 35 2425588
[44] Takeuchi A and Inoue A 2005 Mater. Trans. 46 2817
[45] Huang B, Yang Y,Wang A D,Wang Q and Liu C T 2017 Intermetallics 84 74
[46] Zhou Z Q, Shang Y H, Liu X D and Yang Y 2023 npj Comput. Mater. 9 15
[47] Long T, Zhang Y, Fortunato M N, Shen C, Dai M and Zhang H B 2022 Acta Mater. 231 117898
[48] Li Z and Birbilis N 2024 npj Comput. Mater. 10 112
[49] Zhang Y X, Xie S J, Guo W, Ding J, Poh L H and Sha Z D 2023 J. Alloy. Compd. 960 170793
[50] Wang Y H, Tian Y F, Kirk T, Laris O, Ross Jr J H, Noebe R D, Keylin V and Arroyave R 2020 Acta Mater. 194 144
[51] Li X, Shan G C, Zhang J L and Sheck C H 2022 J. Mater. Chem. C 10 17291
[52] Li X W, Chang L, Cao Y, Lu J Q, Lu X L and Jiang H Q 2023 Proc. Natl. Acad. Sci. USA 120 e2309062120
[53] Lundberg S M and Lee S I 2017 arXiv:1705.07874 [cs.AI]
[54] Han Y, Ding J, Kong F L, Inoue A, Zhu S L, Wang Z, Shalaan E, Al- Marzouki F 2017 J. Alloy. Compd. 691 364
[55] Williams A, Moruzzi V, Malozemoff A and Terakura K 1983 IEEE Trans. Magn. 19 1983
[56] Zang B, Parsons R, Onodera K, Kishimoto H, Kato A, Liu A C Y and Suzuki K 2017 Scr. Mater. 132 68
[57] Suzuki K, Parsons R, Zang B, Onodera K, Kishimoto H, Shoji T and Kato A 2018 J. Alloy. Compd. 735 613
[58] Xie L, Liu T, He A, Li Q, Gao Z K, Wang A, Chang C T and Wang X X 2017 J. Mater. Sci. 53 1437
[59] Hou L, Fan X,Wang Q Q, YangWM and Shen B L 2019 J. Mater. Sci. Technol. 35 1655
[60] Naohara T 1996 Metall. Mater. Trans. A 27 3424
[61] Xu J, Yang Y Z, Li W and Chen X C 2016 J. Non-Cryst. Solids 447 167
[62] Zhang X Y, Huang X Y, Liang Y W, Cai Y F, Chen Y A, Gao M, Kawazoe Y, Umetsu R, Xiang M L, Zhao X J, Wang Y C, Wang J Q and Zhang Y 2024 J. Alloy. Compd. 1005 176172

[1]	Inverse design of 3D integrated high-efficiency grating couplers using deep learning Yu Wang(王玉), Yue Wang(王越), Guohui Yang(杨国辉), Kuang Zhang(张狂), Xing Yang(杨星), Chunhui Wang(王春晖), and Yu Zhang(张雨). Chin. Phys. B, 2026, 35(2): 024101.
[2]	Deep learning-enabled inverse design of polarization-selective structural color based on coding metasurface Haolin Yang(杨昊霖), Bo Ni(倪波), Junhong Guo(郭俊宏), Hua Zhou(周华), and Jianhua Chang(常建华). Chin. Phys. B, 2025, 34(5): 050702.
[3]	A graph neural network approach to the inverse design for thermal transparency with periodic interparticle system Bin Liu(刘斌) and Yixi Wang(王译浠). Chin. Phys. B, 2024, 33(8): 084401.
[4]	Inverse design of nonlinear phononic crystal configurations based on multi-label classification learning neural networks Kunqi Huang(黄坤琦), Yiran Lin(林懿然), Yun Lai(赖耘), and Xiaozhou Liu(刘晓宙). Chin. Phys. B, 2024, 33(10): 104301.
[5]	Ab initio molecular dynamics simulations of nano-crystallization of Fe-based amorphous alloys with early transition metals Yao-Cen Wang(汪姚岑), Yan Zhang(张岩), Yoshiyuki Kawazoe, Jun Shen(沈军), Chong-De Cao(曹崇德). Chin. Phys. B, 2018, 27(11): 116401.
[6]	Serrated magnetic properties in metallic glass by thermal cycle Myong-Chol Ri(李明哲), Sajad Sohrabi, Da-Wei Ding(丁大伟), Bang-Shao Dong(董帮少), Shao-Xiong Zhou(周少雄), Wei-Hua Wang(汪卫华). Chin. Phys. B, 2017, 26(6): 066101.
[7]	Inverse design-based metamaterial transparent device and its multilayer realization Li Ting-Hua (李廷华), Huang Ming (黄铭), Yang Jing-Jing (杨晶晶), Yuan Gang (袁刚), Cai Guang-Hui (蔡光卉). Chin. Phys. B, 2014, 23(5): 054102.

No Suggested Reading articles found!

Viewed

Full text

Abstract

Cited

Metrics
Related Articles

Multi-task deep-learning optimization of trade-off properties for superior-performance Fe-based soft magnetic alloys

Cite this article:

Online attention