other · Figshare · 2015

Classification performance.

Kay H. Brodersen (216270), Thomas M. Schofield (216275), Alexander P. Leff (216279), Cheng Soon Ong (216284), Ekaterina I. Lomakina (216288), Joachim M. Buhmann (159866), Klaas E. Stephan (216293)

Cited by 0Open access

Abstract

<p>Classification based on generative embedding using the model shown in <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1002079#pcbi-1002079-g003" target="_blank">Figure 3</a> was compared to ten alternative methods: anatomical feature selection, contrast feature selection, searchlight feature selection, PCA-based dimensionality reduction, regional correlations based on region means, regional correlations based on eigenvariates, regional z-transformed correlations based on eigenvariates, as well as generative embedding using three biologically unlikely alternative models (see inset legends for abbreviations). (a) The balanced accuracy and its central 95% posterior probability interval show that all methods performed significantly better than chance (50%) with the exception of classification with anatomical feature selection and generative embedding using a nonsensical model. Differences between activation-based methods (light grey) and correlation-based methods (dark grey) were largely statistically indistinguishable. By contrast, using the full model shown in <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1002079#pcbi-1002079-g003" target="_blank">Figure 3</a>, generative embedding (blue) significantly outperformed all other methods, except when used with biologically unlikely models (<a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1002079#pcbi-1002079-g005" target="_blank">Figure 5</a>). (b) Receiver-operating characteristic (ROC) curves of the eleven methods illustrate the trade-off between true positive rate (sensitivity) and false positive rate (1 – specificity) across the entire range of detection thresholds. A larger area under the curve is better. (c) Precision-recall (PR) curves illustrate the trade-off between positive prediction value (precision) and true positive rate (recall). A larger area under the curve is better. Smooth ROC and PR curves were obtained using a binormal assumption on the underlying decision values <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1002079#pcbi.1002079-Brodersen3" target="_blank">[97]</a>. For a numerical summary of all results, see <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1002079#pcbi-1002079-t002" target="_blank">Table 2</a>.</p>

Cite this paper

APA 7

(216270), K. H. B., (216275), T. M. S., (216279), A. P. L., (216284), C. S. O., (216288), E. I. L., (159866), J. M. B., & (216293), K. E. S. (2015). Classification performance. Figshare. https://doi.org/10.1371/journal.pcbi.1002079.g006

MLA 9

(216270), Kay H. Brodersen, et al. “Classification performance.” Figshare, 2015, https://doi.org/10.1371/journal.pcbi.1002079.g006.

Chicago

(216270), Kay H. Brodersen, Thomas M. Schofield (216275), Alexander P. Leff (216279), Cheng Soon Ong (216284), Ekaterina I. Lomakina (216288), Joachim M. Buhmann (159866), and Klaas E. Stephan (216293). 2015. “Classification performance.” Figshare. https://doi.org/10.1371/journal.pcbi.1002079.g006.

Harvard

(216270), K.H.B. et al. (2015) ‘Classification performance’, Figshare. Available at: https://doi.org/10.1371/journal.pcbi.1002079.g006.

BibTeX

@misc{2162702015classification,
  title = {{Classification performance.}},
  author = {(216270), Kay H. Brodersen and (216275), Thomas M. Schofield and (216279), Alexander P. Leff and (216284), Cheng Soon Ong and (216288), Ekaterina I. Lomakina and (159866), Joachim M. Buhmann and (216293), Klaas E. Stephan},
  journal = {Figshare},
  year = {2015},
  doi = {10.1371/journal.pcbi.1002079.g006}
}

Made from the index record. Check names, capitalisation and page numbers against the paper. Need another style? AI for Research has over 10,000.