Lazarus
Eli
Lazarus
Eli
No Thumbnail Available
Search Results
Now showing
1 - 2 of 2
-
ArticleLabeling poststorm coastal imagery for machine learning: measurement of interrater agreement(American Geophysical Union, 2021-09-03) Goldstein, Evan B. ; Buscombe, Daniel ; Lazarus, Eli ; Mohanty, Somya D. ; Rafique, Shah Nafis ; Anarde, Katherine A. ; Ashton, Andrew D. ; Beuzen, Tomas ; Castagno, Katherine ; Cohn, Nicholas ; Conlin, Matthew P. ; Ellenson, Ashley ; Gillen, Megan N. ; Hovenga, Paige A. ; Over, Jin-Si ; Palermo, Rose V. ; Ratliff, Katherine M. ; Reeves, Ian R. B. ; Sanborn, Lily H. ; Straub, Jessamin A. ; Taylor, Luke A. ; Wallace, Elizabeth J. ; Warrick, Jonathan ; Wernette, Phillipe ; Williams, Hannah E.Classifying images using supervised machine learning (ML) relies on labeled training data—classes or text descriptions, for example, associated with each image. Data-driven models are only as good as the data used for training, and this points to the importance of high-quality labeled data for developing a ML model that has predictive skill. Labeling data is typically a time-consuming, manual process. Here, we investigate the process of labeling data, with a specific focus on coastal aerial imagery captured in the wake of hurricanes that affected the Atlantic and Gulf Coasts of the United States. The imagery data set is a rich observational record of storm impacts and coastal change, but the imagery requires labeling to render that information accessible. We created an online interface that served labelers a stream of images and a fixed set of questions. A total of 1,600 images were labeled by at least two or as many as seven coastal scientists. We used the resulting data set to investigate interrater agreement: the extent to which labelers labeled each image similarly. Interrater agreement scores, assessed with percent agreement and Krippendorff's alpha, are higher when the questions posed to labelers are relatively simple, when the labelers are provided with a user manual, and when images are smaller. Experiments in interrater agreement point toward the benefit of multiple labelers for understanding the uncertainty in labeling data for machine learning research.
-
ArticleCumulative versus transient shoreline change : dependencies on temporal and spatial scale(American Geophysical Union, 2011-05-21) Lazarus, Eli ; Ashton, Andrew D. ; Murray, A. Brad ; Tebbens, Sarah ; Burroughs, StephenUsing shoreline change measurements of two oceanside reaches of the North Carolina Outer Banks, USA, we explore an existing premise that shoreline change on a sandy coast is a self-affine signal, wherein patterns of change are scale invariant. Wavelet analysis confirms that the mean variance (spectral power) of shoreline change can be approximated by a power law at alongshore scales from tens of meters up to ∼4–8 km. However, the possibility of a power law relationship does not necessarily reveal a unifying, scale-free, dominant process, and deviations from power law scaling at scales of kilometers to tens of kilometers may suggest further insights into shoreline change processes. Specifically, the maximum of the variance in shoreline change and the scale at which that maximum occurs both increase when shoreline change is measured over longer time scales. This suggests a temporal control on the magnitude of change possible at a given spatial scale and, by extension, that aggregation of shoreline change over time is an important component of large-scale shifts in shoreline position. We also find a consistent difference in variance magnitude between the two survey reaches at large spatial scales, which may be related to differences in oceanographic forcing conditions or may involve hydrodynamic interactions with nearshore geologic bathymetric structures. Overall, the findings suggest that shoreline change at small spatial scales (less than kilometers) does not represent a peak in the shoreline change signal and that change at larger spatial scales dominates the signal, emphasizing the need for studies that target long-term, large-scale shoreline change.