/PaperNotebooks

Primary LanguageJupyter Notebook

This repository contains supplementary notebooks for different papers

Note: For large notebooks which github can not open, please download the script (.ipynb) and open the file in Google colab to see the scripts and analysis

Paper: Biswas, A.; Ziatdinov, M.; Kalinin, S. V. Combining Variational Autoencoders and Physical Bias for Improved Microscopy Data Analysis. arXiv February 8, 2023. https://doi.org/10.48550/arXiv.2302.04216

Notebook/Code link:

Script for custom ELBO loss:

https://github.com/arpanbiswas52/PaperNotebooks/blob/main/custom_traceELBO.py

Script for custom SVI class and trainner class:

https://github.com/arpanbiswas52/PaperNotebooks/blob/main/customsvi.py

https://github.com/arpanbiswas52/PaperNotebooks/blob/main/customsvitrainer.py

Example: STEM of Sm-doped BiFeO3

standard sh-VAE training (without implementing physical constraint to maximize the smoothness of latent space)

Distorted 2D latent manifolds

image

standard sh-VAE training (without implementing physical constraint to maximize the smoothness of latent space)

No distortion

image


Paper: Biswas, A., Rama Vasudevan, Maxim Ziatdinov, Sergei V. Kalinin " Optimizing Training Trajectories in Variational Autoencoders via Latent Bayesian Optimization Approach" 2023 Mach. Learn.: Sci. Technol. 4 015011 https://doi.org/10.1088/2632-2153/acb316

Notebook/Code link:

** Notebook for MNIST dataset:**

https://github.com/arpanbiswas52/PaperNotebooks/blob/main/LatentBO_jrVAE_MNIST(Notebook).ipynb

** Notebook for experimental plasmonic nanoparticles dataset:**

https://github.com/arpanbiswas52/PaperNotebooks/blob/main/LatentBO_jrVAE_PlasmonicNano(Notebook).ipynb

Example: MNIST

suboptimal tuning of joint-rotationally invariant variational autoencoder

Overlapping of digits classifications in 2D latent manifolds

image

optimal tuning of joint-rotationally invariant variational autoencoder through latent Bayesian optimization

classification and disentanglement of all digits in 2D latent manifolds

image


Paper: Arpan Biswas, Anna N. Morozovska, Maxim Ziatdinov, Eugene A. Eliseev, and Sergei V. Kalinin “Multi-objective Bayesian optimization of ferroelectric materials with interfacial control for memory and energy storage applications” Journal of Applied Physics 130, 204102 (2021); https://doi.org/10.1063/5.0068903

Notebook/Code link:

https://github.com/arpanbiswas52/MOBO_AFI_Supplements

Example (PZO): Exploration over parameter space containing FE and AFE hysteresis loops:

Building Pareto front to maximize energy storage and loss with Multi-objective Bayesian optimization

image


Paper: Anna N. Morozovska, Eugene A. Eliseev, Arpan Biswas, Hanna V. Shevliakova, Nicholas V. Morozovsky, Sergei V. Kalinin “Chemical control of polarization in thin strained films of a multiaxial ferroelectric: phase diagrams and polarization rotation” Phys. Rev. B 105, 094112 – Published 23 March 2022; https://doi.org/10.1103/PhysRevB.105.094112

Notebook/Code link:

https://github.com/arpanbiswas52/PaperNotebooks/blob/main/BTO_GPexplore.ipynb

Paper: Anna N. Morozovska, Eugene A. Eliseev, Arpan Biswas, Nicholas V. Morozovsky, and Sergei V. Kalinin “Effect of Surface Ionic Screening on Polarization Reversal and Phase Diagrams in Thin Antiferroelectric Films for Information and Energy Storage” Phys. Rev. Applied 16, 044053 – Published 27 October 2021; https://doi.org/10.1103/PhysRevApplied.16.044053

Notebook/Code link:

https://github.com/arpanbiswas52/PaperNotebooks/blob/main/films_PZO_GPexplore.ipynb


Please feel free to use these notebooks. Please cite the relevant papers and email at arpanbiswas52@gmail.com or biswasa@ornl.gov for any further questions