Why domain scientists know more ML than they think — Maxim Ziatdinov¹ ² & Sergei V. Kalinin¹ ¹ Center for Nanophase Materials Sciences and ² Computational Sciences and Engineering Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, United States About 20 years ago, Donald Rumsfeld, the erstwhile minister of defense, famously said referring to the intelligence situation, “…there are…

Maxim Ziatdinov¹ ² & Sergei V. Kalinin¹ ¹ Center for Nanophase Materials Sciences and ² Computational Sciences and Engineering Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, United States For many scientists and engineers, the road towards their profession started from microscopy. Under a simple optical microscope or a…

Maxim Ziatdinov¹ ² & Sergei V. Kalinin¹ ¹ Center for Nanophase Materials Sciences and ² Computational Sciences and Engineering Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, United States Despite the extreme disparity in terms of objects and study methods, some tasks are common across multiple scientific fields. One…

How variational autoencoders can be used to analyze one-dimensional signals — Maxim Ziatdinov¹ ² & Sergei V. Kalinin¹ ¹ Center for Nanophase Materials Sciences and ² Computational Sciences and Engineering Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, United States Data often comes in the form of one-dimensional signals. These can be the time series recording from a sensor or…

Introduction to joint (rotationally-invariant) VAEs that can perform unsupervised classification and disentangle relevant (continuous) factors of variation at the same time. — Maxim Ziatdinov¹ ² & Sergei V. Kalinin¹ ¹ Center for Nanophase Materials Sciences and ² Computational Sciences and Engineering Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, United States What are joint (rotationally-invariant) variational autoencoders, and why would we need them? The short answer to the first question is…

Introduction to unsupervised and class-conditioned variational autoencoders (VAEs) with rotational invariance and their application to image analysis — Maxim Ziatdinov¹ ² & Sergei V. Kalinin¹ ¹ Center for Nanophase Materials Sciences and ² Computational Sciences and Engineering Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, United States Scientific research often yields enormous volumes of data. The examples range from fields such as astronomy with its exquisite optical…