Beecroft Institute for Particle Astrophysics and Cosmology

Abstract:

Citizen science has the potential to expand the scope and scale of research in ecology and conservation, but many professional researchers remain skeptical of data produced by nonexperts. We devised an approach for producing accurate, reliable data from untrained, nonexpert volunteers. On the citizen science website www.snapshotserengeti.org, more than 28,000 volunteers classified 1.51 million images taken in a large-scale camera-trap survey in Serengeti National Park, Tanzania. Each image was circulated to, on average, 27 volunteers, and their classifications were aggregated using a simple plurality algorithm. We validated the aggregated answers against a data set of 3829 images verified by experts and calculated 3 certainty metrics—level of agreement among classifications (evenness), fraction of classifications supporting the aggregated answer (fraction support), and fraction of classifiers who reported “nothing here” for an image that was ultimately classified as containing an animal (fraction blank)—to measure confidence that an aggregated answer was correct. Overall, aggregated volunteer answers agreed with the expert-verified data on 98% of images, but accuracy differed by species commonness such that rare species had higher rates of false positives and false negatives. Easily calculated analysis of variance and post-hoc Tukey tests indicated that the certainty metrics were significant indicators of whether each image was correctly classified or classifiable. Thus, the certainty metrics can be used to identify images for expert review. Bootstrapping analyses further indicated that 90% of images were correctly classified with just 5 volunteers per image. Species classifications based on the plurality vote of multiple citizen scientists can provide a reliable foundation for large-scale monitoring of African wildlife.

Abstract:

We investigate the development of scientific content knowledge of volunteers participating in online citizen science projects in the Zooniverse (www.zooniverse.org). We use econometric methods to test how measures of project participation relate to success in a science quiz, controlling for factors known to correlate with scientific knowledge. Citizen scientists believe they are learning about both the content and processes of science through their participation. We don’t directly test the latter, but we find evidence to support the former - that more actively engaged participants perform better in a project-specific science knowledge quiz, even after controlling for their general science knowledge. We interpret this as evidence of learning of science content inspired by participation in online citizen science.

Abstract:

We predict the evolution of the radio continuum sky at 1.4 GHz from the Horizon-AGN Adaptive Mesh Refinement (AMR) cosmological hydrodynamical simulation of a cubic volume of the Universe 100h−1 Mpc on a side. With empirically motivated models for the radio continuum emission due to both star formation and Active Galactic Nuclei (AGN), we estimate the contribution of each of these processes to the local radio continuum luminosity function (LF) and describe its evolution up to redshift 4. Despite the simplicity of these models, we find that our predictions for the local luminosity function are fairly consistent with Mauch & Sadler (2007) observations, with the faint end of the luminosity function dominated by star forming galaxies and the bright end by radio loud AGNs. At redshift one, a decent match to Smolcic et al. (2009) VLA data in the COSMOS field can only be achieved when we account for radio continuum emission from AGNs. We predict that the strongest evolution across the peak epoch of cosmic activity happens for low luminosity star forming galaxies L1.4GHz < 1022 W Hz−1 , whose contribution rises until z ∼ 2 and declines at higher redshifts. The contribution of low luminosity AGNs L1.4GHz < 1022 W Hz−1 steadily declines from z = 0 throughout the redshift range, whilst that of radio loud objects with luminosities in the range 1022 W Hz−1 < L1.4GHz < 1024 W Hz−1 rises dramatically until z = 4. Finally, high-luminosity radio loud AGNs, with L1.4GHz > 1024 W Hz−1 show surprisingly little evolution from z = 0 to z = 4.