Figure from Baron & Ménard (2020), where we presented the Sequencer algorithm. This algorithm is designed to automatically detect sequences in datasets. If a sequence exists, the Sequencer reorders the objects in the sample according to the detected sequence. 

The left panel of the figure shows an input dataset, where each row represents a one-dimensional object, and each pixel is color-coded according to the intensity in this dimension. The right panel shows the Sequencer output. The Sequencer detected a significant one-dimensional sequence (Einstein's face) and reordered the data accordingly. 
In my research, I develop novel techniques to perform unsupervised anomaly detection and dimensionality reduction of large and complex datasets. The application of these tools has already led to several new discoveries in astronomy (see here and here) and geology (see here).