Decoding the shift-invariant data: applications for band-excitation scanning probe microscopy

TL;DR

This paper introduces a shift-invariant variational autoencoder (shift-VAE) for unsupervised analysis of spectral data, effectively disentangling physically relevant shifts from other variables across various microscopy and spectroscopy applications.

Contribution

The paper presents a novel shift-VAE model that can analyze spectral data with shifts, extending its application to band-excitation microscopy and other spectroscopic techniques.

Findings

Shift-VAE accurately matches ground truth parameters on synthetic data.

It effectively disentangles resonance shifts from peak shape parameters in BE-PFM data.

The approach enables denoising and dimensionality reduction in spectral imaging.

Abstract

A shift-invariant variational autoencoder (shift-VAE) is developed as an unsupervised method for the analysis of spectral data in the presence of shifts along the parameter axis, disentangling the physically-relevant shifts from other latent variables. Using synthetic data sets, we show that the shift-VAE latent variables closely match the ground truth parameters. The shift VAE is extended towards the analysis of band-excitation piezoresponse force microscopy (BE-PFM) data, disentangling the resonance frequency shifts from the peak shape parameters in a model-free unsupervised manner. The extensions of this approach towards denoising of data and model-free dimensionality reduction in imaging and spectroscopic data are further demonstrated. This approach is universal and can also be extended to analysis of X-ray diffraction, photoluminescence, Raman spectra, and other data sets.