We've made progress towards stable and scalable training of energy-based models
[http://yann.lecun.com/exdb/publis/pdf/lecun-06.pdf] (EBMs) resulting in better
sample quality and generalization ability than existing models. Generation in
EBMs spends more compute to continually refine its answers and doing so can
generate samples competitive with GANs [https://arxiv.org/abs/1406.2661] at low
temperatures[1], while also having mode coverage guarantees of likelihood-based
models [https://arxiv.or
|
|
Scooped by
Scott Turner
onto artificial intelligence for students March 22, 2019 1:41 PM
|
Your new post is loading...
