Band Alignment of Oxides by Learnable Structural-Descriptor-Aided Neural Network and Transfer Learning

Kiyohara, Shin; Hinuma, Yoyo; Oba, Fumiyasu

doi:10.1021/jacs.3c13574

Cited by 3 publications

(1 citation statement)

References 88 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Similarity, Li et al accurately predict the formation energy of perovskite oxides by training on 5329 spinel oxides and finetuning on 855 perovskite oxides 17 . However, current transfer learning applications are either between different properties with the same materials (cross-property) or between different materials with the same property (cross-material) 18 – 25 . This is owing to that the effectiveness of transfer learning is closely related to the difference between the source and target domain, and if the domain difference is too large, it will not be effective and may give poorer predictions, i.e., negative transfer 26 .…”

Section: Introductionmentioning

confidence: 99%

From bulk effective mass to 2D carrier mobility accurate prediction via adversarial transfer learning

Chen,

Lu,

Chen

et al. 2024

Nat Commun

View full text Add to dashboard Cite

Data scarcity is one of the critical bottlenecks to utilizing machine learning in material discovery. Transfer learning can use existing big data to assist property prediction on small data sets, but the premise is that there must be a strong correlation between large and small data sets. To extend its applicability in scenarios with different properties and materials, here we develop a hybrid framework combining adversarial transfer learning and expert knowledge, which enables the direct prediction of carrier mobility of two-dimensional (2D) materials using the knowledge learned from bulk effective mass. Specifically, adversarial training ensures that only common knowledge between bulk and 2D materials is extracted while expert knowledge is incorporated to further improve the prediction accuracy and generalizability. Successfully, 2D carrier mobilities are predicted with the accuracy over 90% from only crystal structure, and 21 2D semiconductors with carrier mobilities far exceeding silicon and suitable bandgap are successfully screened out. This work enables transfer learning in simultaneous cross-property and cross-material scenarios, providing an effective tool to predict intricate material properties with limited data.

show abstract

Section: Introductionmentioning

confidence: 99%

From bulk effective mass to 2D carrier mobility accurate prediction via adversarial transfer learning

Chen,

Lu,

Chen

et al. 2024

Nat Commun

View full text Add to dashboard Cite

show abstract

A surface emphasized multi-task learning framework for surface property predictions: A case study of magnesium intermetallics

Shi,

Wang,

Yang

et al. 2024

Journal of Magnesium and Alloys

View full text Add to dashboard Cite

Theoretical and data-driven approaches to semiconductors and dielectrics: from prediction to experiment

Oba,

Nagai,

Katsube

et al. 2024

Science and Technology of Advanced Materials

View full text Add to dashboard Cite

Computational approaches using theoretical calculations and data scientific methods have become increasingly important in materials science and technology, with the development of relevant methodologies and algorithms, the availability of large materials data, and the enhancement of computer performance. As reviewed herein, we have developed computational methods for the design and prediction of inorganic materials with a particular focus on the exploration of semiconductors and dielectrics. High-throughput first-principles calculations are used to systematically and accurately predict the local atomic and electronic structures of polarons, point defects, surfaces, and interfaces, as well as bulk fundamental properties. Machine learning techniques are utilized to efficiently predict various material properties, construct phase diagrams, and search for materials satisfying target properties. These computational approaches have elucidated the mechanisms behind material functionalities and explored promising materials in combination with synthesis, characterization, and device fabrication. Examples include the development of ternary nitride semiconductors for potential optoelectronic and photovoltaic applications, the exploration of phosphide semiconductors and the optimization of heterointerfaces toward the improvement of phosphide-based photovoltaic cells, and the discovery of ferroelectricity in layered perovskite oxides and the theoretical understanding of its origin, all of which demonstrate the effectiveness of our computer-aided materials research.

show abstract

Band Alignment of Oxides by Learnable Structural-Descriptor-Aided Neural Network and Transfer Learning

Cited by 3 publications

References 88 publications

From bulk effective mass to 2D carrier mobility accurate prediction via adversarial transfer learning

From bulk effective mass to 2D carrier mobility accurate prediction via adversarial transfer learning

A surface emphasized multi-task learning framework for surface property predictions: A case study of magnesium intermetallics

Theoretical and data-driven approaches to semiconductors and dielectrics: from prediction to experiment

Contact Info

Product

Resources

About