search

marco-rudolph / differnet

This is the official repository to the WACV 2021 paper "Same Same But DifferNet: Semi-Supervised Defect Detection with Normalizing Flows" by Marco Rudolph, Bastian Wandt and Bodo Rosenhahn.
215 stars 67 forks source link

Why is the loss function negative #40

Closed TaoTXiXi closed 1 year ago

TaoTXiXi commented 2 years ago

Why is the loss function calculated negative in the training process

marco-rudolph commented 1 year ago

As the loss reflects the negative log likelihood, it can be negative as the log likelihood can be positive.