Clustering superconductors using unsupervised machine learning

Roter, B.; Ninkovic, N.; Dordevic, S. V.

doi:10.1016/j.physc.2022.1354078

Cited by 25 publications

(22 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…SuperCon is the largest database for superconducting materials with around 30 000 superconductors before filtering. Similar to what was done in some previous studies [6,15], we only used the chemical compositions of the materials extracted from SuperCon. On the other hand, OQMD is a much larger database with around 10 6 DFT-calculated compounds, most of which are not superconductors.…”

Section: Data Collectionmentioning

confidence: 99%

See 1 more Smart Citation

ScGAN: a generative adversarial network to predict hypothetical superconductors

Kim,

Dordevic

2023

J. Phys.: Condens. Matter

View full text Add to dashboard Cite

Despite having been discovered more than three decades ago, high temperature superconductors (HTSs) lack both an explanation for their mechanisms and a systematic way to search for them. To aid this search, this project proposes ScGAN, a generative adversarial network (GAN) to efficiently predict new superconductors. ScGAN was trained on compounds in Open Quantum Materials Database and then transfer learned onto the SuperCon database or a subset of it. Once trained, the GAN was used to predict superconducting candidates, and approximately 70% of them were determined to be superconducting by a classification model–a 23-fold increase in discovery rate compared to manual search methods. Furthermore, more than 99% of predictions were novel materials, demonstrating that ScGAN was able to potentially predict completely new superconductors, including several promising HTS candidates. This project presents a novel, efficient way to search for new superconductors, which may be used in technological applications or provide insight into the unsolved problem of high temperature superconductivity.

show abstract

Section: Data Collectionmentioning

confidence: 99%

“…Clustering is an unsupervised machine learning technique whose main goal is to unveil hidden patterns in the data. It was recently applied to superconductors from SuperCon database [15]. Depending on the data set, different clustering algorithms were used, such as k-means, hierarchical, Gaussian mixtures, self-organizing maps, etc.…”

Section: Clusteringmentioning

confidence: 99%

ScGAN: a generative adversarial network to predict hypothetical superconductors

Kim,

Dordevic

2023

J. Phys.: Condens. Matter

View full text Add to dashboard Cite

show abstract

“…As mentioned above, and in line with others in the literature [13,14,15,16,17], we adopted a convenient source of data, namely the SuperCon database [18] which collects the values of critical temperatures T c for superconducting materials known from literature. To our knowledge, SuperCon represents the largest database of its kind, from which we have extracted a list of ∼ 16, 000 materials.…”

Section: Resultsmentioning

confidence: 99%

“…Le et al [15] trained and validated a Variational Bayesian Neural Network using superconductors compositionbased features for the T c prediction. Roter et al used only chemical elements and stoichiometry, with no extracted features, to predict the critical temperature [16] and to cluster superconductors [17].…”

Section: Introductionmentioning

confidence: 99%

Leveraging Composition-Based Energy Material Descriptors for Machine Learning Models

Trezza¹,

Chiavazzo

2023

Preprint

View full text Add to dashboard Cite

“…Machine learning (ML) has become an essential part of materials design and discovery, − providing the ability to develop the form of structure/property relationships and solve them simultaneously. Supervised ML can predict the properties or classes of materials (the labels) based on the physicochemical characteristics (the features) − or even the required features based on the desired properties or product specifications using inverse design. − Using unsupervised learning, it is also possible to identify hidden patterns, trends, and relationships among different materials based on their similarities in a high-dimensional feature space, regardless of their functional properties. − This can be very useful to discover new classes of materials, identify the pure archetypes or representative prototypes, , or visualize the distribution of high-dimensional data to rapidly assess the likelihood of structure/property relationships worth exploring . Successful use of ML assumes, however, that sufficient machine-readable data is available.…”

Section: Introductionmentioning

confidence: 99%

Structure-Free Mendeleev Encodings of Material Compounds for Machine Learning

Zhuang,

Barnard

2023

Chem. Mater.

View full text Add to dashboard Cite

Machine learning is a powerful tool to predict the properties of materials for a variety of applications. However, generating data sets of carefully characterized materials can be time-consuming and costly, particularly when numerous candidate materials are later found to be irrelevant. The problem could be alleviated if machine learning can be used with minimal information to provide guidance at an early stage before significant investment has been made. Since structural characterization is one of the most expensive parts of the process, this study explores structure-free encoding of materials using Mendeleev encoding, a method that does not require information such as lattice constants, lattice positions, or bonding networks. We evaluate Mendeleev encoding using three data sets of continuous, complex material compounds used for battery applications, with four different unsupervised learning methods, inclusive of six algorithms and four evaluation metrics and in addition visualizations of the results. Our results show that Mendeleev encoding is more accurate, stable, and reliable than alternative structure-free encoding, allowing both principle component analysis and archetypal analysis to capture more of the variance during dimensionality reduction and consistently provide superior clustering results. Mendeleev encoding is a simple and scientifically intuitive way of representing material data that is both human and machine-readable and is applicable to any machine-learning task training with tabular data.

show abstract

Clustering superconductors using unsupervised machine learning

Cited by 25 publications

References 14 publications

ScGAN: a generative adversarial network to predict hypothetical superconductors

ScGAN: a generative adversarial network to predict hypothetical superconductors

Leveraging Composition-Based Energy Material Descriptors for Machine Learning Models

Structure-Free Mendeleev Encodings of Material Compounds for Machine Learning

Contact Info

Product

Resources

About