Open Access

Table 5

Peak identification quality metrics for approximation models applied to the PLAsTiCC.

Model RMSE, days MAE, days RSE RAE MAPE, %
GP 1.76 ± 0.07 1.22 ± 0.04 0.0057 ± 0.0002 0.0045 ± 0.0002 0.00202 ± 0.00006
MLP (sklearn) 1.67 ± 0.06 1.15 ± 0.04 0.0054 ± 0.0002 0.0042 ± 0.0002 0.00192 ± 0.00006
MLP (pytorch) 1.9 ± 0.1 1.26 ± 0.04 0.0061 ± 0.0004 0.0047 ± 0.0002 0.00210 ± 0.00007
BNN 1.99 ± 0.08 1.34 ± 0.04 0.0064 ± 0.0003 0.0049 ± 0.0002 0.00223 ± 0.00009
NF 2.17 ± 0.09 1.51 ± 0.05 0.0070 ± 0.0003 0.0056 ± 0.0002 0.00251 ± 0.00008

Notes. CNN-based approach is when CNN is trained to predict the true peak position.

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.