Do experts make mistakes? A comparison of human and machine identification of dinoflagellates

Abstract

The authors present evidence of the difficulties facing human taxonomists/ecologists in identifying marine dinoflagellates. This is especially important for work on harmful algal blooms in marine aquaculture. It is shown that it is difficult for people to categorise specimens from species with significant morphological variation, perhaps with morphologies overlapping with those of other species. Trained personnel can be expected to achieve 67 to 83% self-consistency and 43% consensus between people in an expert taxonomic labelling task. Experts who are routinely engaged in particular discriminations can return accuracies in the range of 84 to 95%. In general, neither human nor machine can be expected to give highly accurate or repeatable labelling of specimens. It is also shown that automation methods can perform as well as humans on these complex categorisations.

Network of evolutionary binary classifiers for classification and retrieval in macroinvertebrate databases

“…In a recent study [7] on a set of river macroinvertebrates, aerage correct classification of 88.2% and 75.3% have been achieved in training and test sets. The observed levels of taxonomical accuracy match levels of human accuracy for other aquatic taxonomic groups [8]. In follow-up work on the same dataset [9], a significant improvement on the classification accuracy was achieved.…”

Validação do método de Sedgwick-Rafter para a quantificação do fitoplâncton

“…Segundo Culverhouse et al (2003), analistas humanos não são bons instrumentos de classificação e contagem. Estes autores relatam que a performance humana em identificar e classificar organismos ao microscópio é afetada por muitos fatores psicológicos, entre eles a capacidade de memorizar, a curto prazo, somente de 5 a 9 itens; fadiga e tédio; as constantes atualizações e modificações na taxonomia do fitoplâncton e tendência em superestimar a contagem pela expectativa prévia em encontrar determinado organismo na amostra.…”

“…Em função disso, comparando identificação e contagem de dinoflagelados marinhos entre analistas especialistas e atuantes na rotina de análise com métodos automatizados, Culverhouse et al (2003) A incerteza está relacionada a um resultado de medição e não ao valor verdadeiro do mensurando que, na prática, não é conhecido. Ela caracteriza um intervalo ou faixa de dispersão, e não um valor pontual.…”

“…Para a quantificação do fitoplâncton há métodos microscópicos (BRANCO, 1986;LAWTON et al, 1999 MCDERMOTT, 2010;EDLER;ELBRÄCHTER, 2010) e métodos moleculares (GALLUZZI;PENNA, 2010;GESCHER;MARRIN III;SCHOLIN, 2010;DIERCKS;. Além destes, há métodos que utilizam fluorescência (BEUTLER et al, 2002;LEBOULANGER et al, 2002;MARŠÁLEK, 2004;GREGOR et al, 2005) e contagem de partículas (SIERACKI; SIERACKI; YENTSCH, 1998;DUBELAAR et al, 1999;DUBELAAR;GERRITZEN, 2000;SHALAPYONOK;OLSON, 2007;THYSSEN et al, 2008).…”

Automated measurements of filamentous cyanobacteria by digital image analysis

“…The system is not intended for concurrent identification of several species, and is thus most suitable for situations where one species is dominating and few morphologically similar species are present. For automated identification of several species, other techniques must be applied (see e.g., Culverhouse et al 2003;Embleton et al 2003).…”