Representations in neural network based empirical potentials

Cubuk, Ekin D.; Malone, Brad D.; Onat, Berk; Waterland, Amos; Kaxiras, Efthimios

doi:10.1063/1.4990503

Cited by 43 publications

(45 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To provide a direct comparison between the present model and potentials obtained with the conventional ANN method 11,12 , we constructed two different Li-Si potentials one with the INN method and the other with the ANN method. Since the SNN method relies on the ANN approach for the construction of pure element NNs, it is subject to the same limitations as ANN at the level of pure-element network construction.…”

Section: Model Construction and Validationmentioning

confidence: 99%

Implanted neural network potentials: Application to Li-Si alloys

et al. 2018

Self Cite

View full text Add to dashboard Cite

Modeling the behavior of materials composed of elements with different bonding and electronic structure character for large spatial and temporal scales and over a large compositional range, is a challenging problem. A case in point are amorphous alloys of Si, a prototypical covalent material, and Li, a prototypical metal, which are being considered as anodes for high-energydensity batteries. To address this challenge, we develop a methodology based on neural networks, that extends the conventional training approach to incorporate pre-trained parts that capture the character of different components, into the overall network; we refer to this model as the "implanted neural network" method. We show that this approach works well for the Si-Li amorphous alloys for a wide range of compositions, giving good results for key quantities like the diffusion coefficients. The method is readily generalizable to more complicated situations that involve two or more different elements.

show abstract

Section: Model Construction and Validationmentioning

confidence: 99%

Implanted neural network potentials: Application to Li-Si alloys

et al. 2018

Self Cite

View full text Add to dashboard Cite

show abstract

“…The importance of the descriptor selection from physical considerations has been observed for a diverse set of materials science applications; [1][2][3][4][5][16][17][18][19][20][21][22][23][24][25][26] however, it is not always possible to find the relevant physical descriptors for the desired application. Furthermore, even if physical descriptors have been identified, they are not always easily accessible.…”

Section: Article Scitationorg/journal/jcpmentioning

confidence: 99%

Screening billions of candidates for solid lithium-ion conductors: A transfer learning approach for small data

Cubuk

Sendek

Reed

2019

The Journal of Chemical Physics

Self Cite

108

101

View full text Add to dashboard Cite

Machine learning (ML) methods have the potential to revolutionize materials design, due to their ability to screen materials efficiently. Unlike other popular applications such as image recognition or language processing, large volumes of data are not available for materials design applications. Here, we first show that a standard learning approach using generic descriptors does not work for small data, unless it is guided by insights from physical equations. We then propose a novel method for transferring such physical insights onto more generic descriptors, allowing us to screen billions of unknown compositions for Li-ion conductivity, a scale which was previously unfeasible. This is accomplished by using the accurate model trained with physical insights to create a large database, on which we train a new ML model using the generic descriptors. Unlike previous applications of ML, this approach allows us to screen materials which have not necessarily been tested before (i.e., not on ICSD or Materials Project). Our method can be applied to any materials design application where a small amount of data is available, combined with high details of physical understanding.

show abstract

“…One of the most exciting and challenging aspects of structure–property (e.g., ML) model development is that these three quantities are coupled and ever changing (Figure ). Large data sets increase the benefit of more complex models, and a less tailored representation will benefit from models such as ANNs that essentially carry out feature engineering of the representation . An individual researcher's answers to the questions I posed largely shape what the right balance is in choosing one of many possible trade‐offs in data, model, and representation (Figure ).…”

Section: The Data Model and Representation Trade‐offmentioning

confidence: 99%

Making machine learning a useful tool in the accelerated discovery of transition metal complexes

Kulik

2019

WIREs Comput Mol Sci

View full text Add to dashboard Cite

As machine learning (ML) has matured, it has opened a new frontier in theoretical and computational chemistry by offering the promise of simultaneous paradigm shifts in accuracy and efficiency. Nowhere is this advance more needed, but also more challenging to achieve, than in the discovery of open-shell transition metal complexes. Here, localized d or f electrons exhibit variable bonding that is challenging to capture even with the most computationally demanding methods. Thus, despite great promise, clear obstacles remain in constructing ML models that can supplement or even replace explicit electronic structure calculations. In this article, I outline the recent advances in building ML models in transition metal chemistry, including the ability to approach sub-kcal/mol accuracy on a range of properties with tailored representations, to discover and enumerate complexes in large chemical spaces, and to reveal opportunities for design through analysis of feature importance. I discuss unique considerations that have been essential to enabling ML in open-shell transition metal chemistry, including (a) the relationship of data set size/diversity, model complexity, and representation choice, (b) the importance of quantitative assessments of both theory and model domain of applicability, and (c) the need to enable autonomous generation of reliable, large data sets both for ML model training and in active learning or discovery contexts. Finally, I summarize the next steps toward making ML a mainstream tool in the accelerated discovery of transition metal complexes.

show abstract

Representations in neural network based empirical potentials

Cited by 43 publications

References 50 publications

Implanted neural network potentials: Application to Li-Si alloys

Implanted neural network potentials: Application to Li-Si alloys

Screening billions of candidates for solid lithium-ion conductors: A transfer learning approach for small data

Making machine learning a useful tool in the accelerated discovery of transition metal complexes

Contact Info

Product

Resources

About