Data Mining in Scanning Probe Microscopy

Scanning probe microscopy instrumentation and techniques for data acquisition have made great leaps over the past decade, producing ever more complex and large multidimensional datasets. This necessitates the development of new instrumentation and techniques for data analysis, in order to better understand this new wealth of data. Fortunately, these developments are paralleled by the general rise of big data and the development of a number of data mining techniques which can extract hidden structure from large datasets. In this talk, I will describe the use of unsupervised learning techniques (e.g. clustering, feature extraction, and spectral mixing) that can identify patterns in data by simultaneously analyzing multiple variables. Using these methods we can uncover statistically significant components and trends within large multidimensional datasets, then compress and highlight these components, allowing us to both better visualize the data and extract physically relevant information.