Data for Concurrent Time-Series Selections Using Deep Learning and Dimension Reduction

This data set has the ground truth labels for the publication "Concurrent Time-Series Selections Using Deep Learning and Dimension Reduction".

The raw tri-axial accelerometry data is available as a separate data set:
dataset doi: 10.5281/zenodo.5500402
original paper for that data set doi: 10.3354/esr00084

The data set here (available at the doi: 10.5281/zenodo.5503031) has the ground truth labels in 4 columns.

The first and second columns are the start and end of the label in normalized coordinates [-1,1] representing the x-axis mouse clicks that created the label.
The third and fourth columns are transformed back to the data x-axis [0,173255].

1D_1 are the labels for surfacing feature
1D_2 are the labels for diving feature

We also provide a recording of one of the participants in the user study.