Fig. 8
From: Microalgae classification using semi-supervised and active learning based on Gaussian mixture models
Evaluation of the different instances of the semi-supervised data. In order to obtain statistical results, we generated ten different instances for each supervision percentage. The accuracy for all ten instances is shown using the proposed method with entropy, in the first two columns, and random selection, in the last column. The visualization is improved using different colors. the vertical axis represents the accuracy and the horizontal axis represents the number of active samples informed to the system. It is important to call attention to the vertical axis, where the intervals are different between the results from the first and the second datasets. a Results for 1 % of semi-supervision in the first dataset. b Results for 1 % of semi-supervision in the second dataset. c Results for 1 % of semi-supervision in the second dataset. d Results for 3 % of semi-supervision in the first dataset. e Results for 3 % of semi-supervision in the second dataset. f Results for 3 % of semi-supervision in the second dataset. g Results for 10 in the first dataset. h Results for 10 % of semi-supervision in the second dataset. i Results for 10 % of semi-supervision in the second dataset (color figure online)