Generative artificial intelligence and its applications in materials science: Current situation and future perspectives

Liu, Yue; Yang, Zhengwei; Yu, Zhenyao; Liu, Zitu; Liu, Dahui; Lin, Hailong; Li, Ming‐Qing; Ma, Shuchang; Avdeev, Maxim; Shi, Siqi

doi:10.1016/j.jmat.2023.05.001

Cited by 91 publications

(31 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, the performance of ML models is significantly contingent upon the quality and diversity of input data. 47 Currently, the utilization of highly reliable experimental data contributes to the construction of more robust models. For classification tasks, the acquisition of labeled data is essential, whereas regression tasks necessitate the gathering of data pertaining to input features and corresponding continuous numerical values (target variables).…”

Section: ■ Introductionmentioning

confidence: 99%

“…The application of ML in materials science has demonstrated remarkable potential, holding promise of expediting the discovery, optimization, and design processes of novel materials. However, the performance of ML models is significantly contingent upon the quality and diversity of input data . Currently, the utilization of highly reliable experimental data contributes to the construction of more robust models.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Exploring the Ambient-Temperature Degradation Reactions of PET through Two-Step Machine Learning and High-Throughput Experimentation

Wang,

Li,

Meng

et al. 2024

ACS Sustainable Chem. Eng.

View full text Add to dashboard Cite

Poly(ethylene terephthalate) (PET) is a prevalent single-use plastic, posing a significant threat to the environment and human health due to its nondegradable properties. To confront this pressing issue, there is an urgent need for a method of PET degradation that is not only efficient but also cost-effective. Recognizing the intricate interactions of parameters, we propose an innovative approach that integrates two-step machine learning (ML) techniques to achieve our objective while minimizing experimental costs. A dataset comprising 120 distinct PET degradation conditions was created using high-throughput experimentation (HTE). Initially, eight ML algorithms were employed to train and predict performance, with the decision tree model yielding the most favorable outcomes. Subsequently, the trained ML model was expanded to encompass an extensive array of hypothetical reaction conditions, facilitating the identification of degradation formulations that exhibit exceptional performance. Additionally, through the integration of feature importance analysis, we systematically reconstruct a relevant chemical space. Following regression training and prediction, we identified reaction conditions with significantly higher degradation rates at ambient temperature. The utilization of these conditions not only enhances the efficacy of research and development, leading to reduced experimental costs, but also provides valuable insights for future investigations into PET degradation. In contrast to conventional, resource-intensive trial-and-error approaches, our established platform facilitates the assessment of PET degradation rates across diverse reaction conditions, enabling a preliminary screening for process optimization. Consequently, this approach contributes to the mitigation of solid waste pollution and the advancement of sustainable economic development.

show abstract

Section: ■ Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Exploring the Ambient-Temperature Degradation Reactions of PET through Two-Step Machine Learning and High-Throughput Experimentation

Wang,

Li,

Meng

et al. 2024

ACS Sustainable Chem. Eng.

View full text Add to dashboard Cite

show abstract

“…Within the realm of AI, third, the GAI methods (e.g., generative adversarial networks (GANs), variational autoencoders) are other emerging tools with outstanding ability to learn complex patterns in the data and generate new contents, properties, or structures. [13] The GAI algorithms are particularly strong for inverse design of material systems. [13] Despite the significant capacity of the GAI methods in enhancing TENG systems in various aspects, their substantive application in this area is conspicuous by its absence.…”

Section: Introductionmentioning

confidence: 99%

“…[13] The GAI algorithms are particularly strong for inverse design of material systems. [13] Despite the significant capacity of the GAI methods in enhancing TENG systems in various aspects, their substantive application in this area is conspicuous by its absence. The major challenge with the implementation of AI and GAI methods is that they often require a large amount of data for calibration.…”

Section: Introductionmentioning

confidence: 99%

Maximizing Triboelectric Nanogenerators by Physics‐Informed AI Inverse Design

Jiao,

Wang,

Alavi

2023

Advanced Materials

View full text Add to dashboard Cite

Triboelectric nanogenerators offer an environmentally friendly approach to harvesting energy from mechanical excitations. This capability has made them widely sought‐after as an efficient, renewable, and sustainable energy source, with the potential to decrease reliance on traditional fossil fuels. However, developing triboelectric nanogenerators with specific output remains a challenge mainly due to the uncertainties associated with their complex designs for real‐life applications. Artificial intelligence‐enabled inverse design is a powerful tool to realize performance‐oriented triboelectric nanogenerators. This is an emerging scientific direction that can address the concerns about the design and optimization of triboelectric nanogenerators leading to a next generation nanogenerator systems. This perspective paper aims at reviewing the principal analysis of triboelectricity, summarizing the current challenges of designing and optimizing triboelectric nanogenerators, and highlighting the physics‐informed inverse design strategies to develop triboelectric nanogenerators. Strategic inverse design is particularly discussed in the contexts of expanding the four‐mode analytical models by physics‐informed artificial intelligence, discovering new conductive and dielectric materials, and optimizing contact interfaces. Various potential development levels of artificial intelligence‐enhanced triboelectric nanogenerators are delineated. Finally, the potential of physics‐informed artificial intelligence inverse design to propel triboelectric nanogenerators from prototypes to multifunctional intelligent systems for real‐life applications is discussed.

show abstract

“…Potential energy surface approaches based on empirical interatomic potentials − are fast, however, often inaccurate. Recently, machine learning (ML) techniques − with density functional theory (DFT) results as inputs yield fast simulations and accurate results. One popular choice is the descriptor-based neural network (NN) approaches ,,− , by regarding the energy as a function of bond lengths, bond angles, or related symmetry functions. ,− ,, However, there are three main issues: (1) large amount of symmetry functions, predetermined subjectively; (2) time-consuming and poor accuracy in force fitting, based on gradient of energy; (3) no objective distribution criteria in the training data generation.…”

mentioning

confidence: 99%

Parameter-Free and Electron Counting Satisfied Material Representation for Machine Learning Potential Energy and Force Fields

Xi,

Chan,

Bao

et al. 2024

J. Phys. Chem. Lett.

View full text Add to dashboard Cite

We proposed a parameter-free volume element representation that satisfies the electron counting model and obtains accurate machine learning potential energy and direct force fitting of randomly perturbed hexagonal BN. Our method preserves permutational, translational, and rotational invariance and can be extended to three-dimensional systems, verified by a system of bulk Si. As a result, we obtained 0.57 meV/atom potential energy root mean squared error (RMSE) and 59 meV/Å force RMSE for perturbed bulk BN systems and 0.43 meV/atom potential energy RMSE and 36 meV/Å force RMSE for perturbed Si systems. In addition, an unbiased perturbation-based data set construction scheme is introduced and a continuous population distribution is obtained with a training data set of 4500, which is about 1 order of magnitude smaller than standard methods based on first-principles molecular dynamics simulations and saves a large amount of computing resources. General validity of our model is verified by structure optimization, molecular dynamics simulations, and extrapolations.

show abstract

Generative artificial intelligence and its applications in materials science: Current situation and future perspectives

Cited by 91 publications

References 35 publications

Exploring the Ambient-Temperature Degradation Reactions of PET through Two-Step Machine Learning and High-Throughput Experimentation

Exploring the Ambient-Temperature Degradation Reactions of PET through Two-Step Machine Learning and High-Throughput Experimentation

Maximizing Triboelectric Nanogenerators by Physics‐Informed AI Inverse Design

Parameter-Free and Electron Counting Satisfied Material Representation for Machine Learning Potential Energy and Force Fields

Contact Info

Product

Resources

About