Open Access

Table A.4

Effect of our model’s stability by different selections of M and K.

Different selections of M and K Precision Accuracy Recall F1 score MCC AP Sum
ZTF-NEWm

M=100, K=900 96.2±6.0 94.9±2.8 87.1±11.9 90.8±5.5 88.0±6.5 99.8±0.3 556.7±16.1
M=200, K=800 96.2±3.4 97.4±1.4 95.2±5.5 95.6±2.5 93.9±3.3 99.7±0.3 578.0±7.8
M=300, K=700 98.3±1.0 98.4±0.4 96.5±1.7 97.4±0.7 96.3±0.9 99.9±0.1 586.7±2.3
M=400, K=600 97.4±0.7 98.5±0.2 97.6±1.1 97.5±0.2 96.4±0.4 99.7±0.1 587.1±1.3
M=500, K=500 97.3±0.9 98.5±0.3 97.8±0.5 97.5±0.5 96.5±0.8 99.9±0.1 587.4±1.5
M=600, K=400 97.6±0.8 98.5±0.5 97.5±1.7 97.5±0.8 96.5±1.2 99.7±0.2 587.4±2.4
M=700, K=300 97.7±0.9 98.5±0.3 97.5±0.8 97.6±0.5 96.5±0.7 99.5±0.4 587.3±1.6
M=800, K=200 97.3±1.1 98.5±0.2 97.7±1.4 97.5±0.4 96.4±0.6 99.6±0.4 587.0±2.0
M=900, K=100 97.6±0.4 98.6±0.1 97.6±0.3 97.6±0.2 96.6±0.3 99.5±0.6 587.5±0.8

Notes. It can be shown that our algorithm’s results tend to stabilize once M ≥ 400. We employ precision, accuracy, recall, F1 score, Matthews correlation coefficient (MCC), and average precision (AP) as performance measurement indicators. The overall performance measurement is obtained by summing the values of these six indicators (shown in the last column). All results listed are SSLS (R = 3).

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.