Estimating False Positive Contamination in Crater Annotations from Citizen Science Data

Abstract

Web-based citizen science often involves the classification of image features by large numbers of minimally trained volunteers, such as the identification of lunar impact craters under the Moon Zoo project. Whilst such approaches facilitate the analysis of large image data sets, the inexperience of users and ambiguity in image content can lead to contamination from false positive identifications. We give an approach, using Linear Poisson Models and image template matching, that can quantify levels of false positive contamination in citizen science Moon Zoo crater annotations. Linear Poisson Models are a form of machine learning which supports predictive error modelling and goodness-of-fits, unlike most alternative machine learning methods. The proposed supervised learning system can reduce the variability in crater counts whilst providing predictive error assessments of estimated quantities of remaining true verses false annotations. In an area of research influenced by human subjectivity, the proposed method provides a level of objectivity through the utilisation of image evidence, guided by candidate crater identifications.

Quantifying biological samples using Linear Poisson Independent Component Analysis for MALDI-ToF mass spectra

“…In cases of a plateau, failing to achieving a true minimum is compensated for in the error theory, as error covariances are scaled by the final goddness-of-fit. Details of the full method and its validation in other applications can be found in Tar and Thacker (2014) and Tar et al (2015 , 2017 ).…”

“…In recent work we derived an ICA method for data with Poisson sampling characteristics, see Tar and Thacker (2014) : Linear Poisson Modelling (LPM). It has been applied to planetary and medical images (see Tar et al , 2015 , 2017 ). We believe this method [(*) in Table 1 ] provides the best match to the properties of MALDI data and is therefore evaluated here on the task of measuring mixtures of complex lipid specimens.…”

The Subjectivity in Identification of Martian Channel Networks and Its Implication for Citizen Science Projects

“…For this to be valid, the public must be able to identify valleys reliably. Although, craters are arguably simpler features to identify than valleys, errors within their identification still arise (Tar et al 2017).…”

The statistical properties of raw and preprocessed ToF mass spectra

“…The aim of our preprocessing is to minimise these effects so that data analysis based upon ideal Poisson statistics can be applied, such as Linear Poisson Modelling. [26][27][28] Preprocessing is an essential step for the analysis of mass spectra, 13 but represents a challenging problem. Poorly performed preprocessing may prevent meaningful analysis of signal.…”