Unsupervised learning of two-component nematicity from STM data on magic angle bilayer graphene

Taranto, William; Lederer, Samuel; Choi, Youngjoon; Izmailov, Pavel; Wilson, Andrew Gordon; Nadj-Perge, Stevan; Kim, Eun-Ah

doi:10.48550/arxiv.2203.04449

Cited by 2 publications

(1 citation statement)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For sufficiently simple systems, the disentangled representations can often be associated with the specific physical factors of variability in the system. In particular, unsupervised ML approaches have allowed the discovery of physics from complex and/or large microscopic images/datasets as in [4][5][6][7][8][9][10][11][12], where the disentangled representations provide insight into specific physical order parameters.…”

Section: Introductionmentioning

confidence: 99%

Optimizing training trajectories in variational autoencoders via latent Bayesian optimization approach ^*

Biswas

Vasudevan²,

Ziatdinov³

et al. 2023

Mach. Learn.: Sci. Technol.

View full text Add to dashboard Cite

Unsupervised and semi-supervised ML methods such as variational autoencoders (VAE) have become widely adopted across multiple areas of physics, chemistry, and materials sciences due to their capability in disentangling representations and ability to find latent manifolds for classification and/or regression of complex experimental data. Like other ML problems, VAEs require hyperparameter tuning, e.g., balancing the Kullback–Leibler (KL) and reconstruction terms. However, the training process and resulting manifold topology and connectivity depend not only on hyperparameters, but also their evolution during training. Because of the inefficiency of exhaustive search in a high-dimensional hyperparameter space for the expensive-to-train models, here we have explored a latent Bayesian optimization (zBO) approach for the hyperparameter trajectory optimization for the unsupervised and semi-supervised ML and demonstrated for joint-VAE with rotational invariances. We have demonstrated an application of this method for finding joint discrete and continuous rotationally invariant representations for MNIST and experimental data of a plasmonic nanoparticles material system. The performance of the proposed approach has been discussed extensively, where it allows for any high dimensional hyperparameter trajectory optimization of other ML models.

show abstract

Section: Introductionmentioning

confidence: 99%

Optimizing training trajectories in variational autoencoders via latent Bayesian optimization approach ^*

Biswas

Vasudevan²,

Ziatdinov³

et al. 2023

Mach. Learn.: Sci. Technol.

View full text Add to dashboard Cite

show abstract

Machine learning the microscopic form of nematic order in twisted double-bilayer graphene

Sobral,

Obernauer,

Turkel

et al. 2023

Nat Commun

View full text Add to dashboard Cite

Modern scanning probe techniques, such as scanning tunneling microscopy, provide access to a large amount of data encoding the underlying physics of quantum matter. In this work, we show how convolutional neural networks can be used to learn effective theoretical models from scanning tunneling microscopy data on correlated moiré superlattices. Moiré systems are particularly well suited for this task as their increased lattice constant provides access to intra-unit-cell physics, while their tunability allows for the collection of high-dimensional data sets from a single sample. Using electronic nematic order in twisted double-bilayer graphene as an example, we show that incorporating correlations between the local density of states at different energies allows convolutional neural networks not only to learn the microscopic nematic order parameter, but also to distinguish it from heterostrain. These results demonstrate that neural networks are a powerful method for investigating the microscopic details of correlated phenomena in moiré systems and beyond.

show abstract

Unsupervised learning of two-component nematicity from STM data on magic angle bilayer graphene

Cited by 2 publications

References 34 publications

Optimizing training trajectories in variational autoencoders via latent Bayesian optimization approach ^*

Optimizing training trajectories in variational autoencoders via latent Bayesian optimization approach ^*

Machine learning the microscopic form of nematic order in twisted double-bilayer graphene

Contact Info

Product

Resources

About

Unsupervised learning of two-component nematicity from STM data on magic angle bilayer graphene

Cited by 2 publications

References 34 publications

Optimizing training trajectories in variational autoencoders via latent Bayesian optimization approach *

Optimizing training trajectories in variational autoencoders via latent Bayesian optimization approach *

Machine learning the microscopic form of nematic order in twisted double-bilayer graphene

Contact Info

Product

Resources

About

Optimizing training trajectories in variational autoencoders via latent Bayesian optimization approach ^*

Optimizing training trajectories in variational autoencoders via latent Bayesian optimization approach ^*