search

hagabbar / matching_cnn_paper

Matching matched filtering with deep learning in gravitational-wave astronomy
2 stars 1 forks source link

Answer confusion matrix SNR 2 conundrum #31

Open chrismessenger opened 6 years ago

chrismessenger commented 6 years ago

Why is the confusion matrix for SNR 2 not uniform?

In fact it seems to have changed again since DCC submission and it looks more uniform now but not quite.

Can we solve this mystery? It is something that we may have to explain since it stands out in the paper.

hagabbar commented 6 years ago

Well the only difference between the new confusion matrix and the one I showed you last week is the mass distribution used in the training/testing of the network. Where the plot last week used the GH masses and the one this week uses your log distribution.