Table 2
Description of different sets used for training and testing.
| Set name | Δttrue | Use in classification | Use in Δt estimation | |||||
|---|---|---|---|---|---|---|---|---|
| set A | U [5.0,24.0] | |||||||
| set B | U [4.0,24.0] | train + test | test only | |||||
| set C | U [3.0,24.0] | |||||||
| set D | U [0.5,30.0] | — | train only | |||||
| set LSST | LSST like | test only | test only | |||||
Notes. Note that, except for set LSST, in the rest of the controlled sets we use µ ~ U[1/3, 3.0] to have a balanced training. Set D is only used to train the model for predicting time delays. Set LSST represents the whole sample of unresolved systems in LSST without any cuts being imposed. We use this set to test our models trained on different sets depending on our purpose of classification or time-delay estimation. Note that we randomly choose between W7 and merger models while generating the light curves for each system (lensed or unlensed) in all the sets.
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.