Abstract | ||
---|---|---|
We present a framework to understand GAN training as alternating density ratio estimation and approximate divergence minimization. This provides an interpretation for the mismatched GAN generator and discriminator objectives often used in practice, and explains the problem of poor sample diversity. We also derive a family of generator objectives that target arbitrary $f$-divergences without minimizing a lower bound, and use them to train generative image models that target either improved sample quality or greater sample diversity. |
Year | Venue | Field |
---|---|---|
2016 | arXiv: Learning | Mathematical optimization,Discriminator,Divergence,Upper and lower bounds,Computer science,Minification,Generative grammar,Density ratio estimation |
DocType | Volume | Citations |
Journal | abs/1612.02780 | 1 |
PageRank | References | Authors |
0.37 | 0 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Ben Poole | 1 | 554 | 52.06 |
Alexander A. Alemi | 2 | 70 | 9.92 |
Jascha Sohl-Dickstein | 3 | 673 | 82.82 |
Anelia Angelova | 4 | 410 | 30.70 |