It is of common knowledge that Supervised Learning uses forward KL divergence. However, I would like to use Reverse KL Divergence and am looking for examples of similar usage in literature. Most importantly, I would like to see how it is implemented. If it is not feasible to implement it, I would also like to know the difficulties and approximations or workarounds to do the same.
Asked
Active
Viewed 40 times
0
-
Reverse KL simply switch the two operands, so you can use any implementation of KL divergence but switch the inputs. – Adam Oudad Jul 08 '20 at 17:20
-
Can you prove that it will work? @AdamOudad – Parth Shah Jul 09 '20 at 11:24