AlexiaJM

4 ms·

1. 1.▲
   Adversarial Score Matching and Consistent Sampling (Better Than SOTA GANs)(ajolicoeur.wordpress.com)

   2 points by AlexiaJM 6y ago|0 comments
2. 2.▲
   by AlexiaJM 7y ago

   Paper is here: https://arxiv.org/abs/1910.06922 . Feel free to ask me questions!
3. 3.▲
   Connections between SVMs, Wasserstein distance and gradient-penalty GANs(ajolicoeur.wordpress.com)

   4 points by AlexiaJM 7y ago|1 comments
4. 4.▲
   On Relativistic F-divergences (AI Research on GANs)(ajolicoeur.wordpress.com)

   3 points by AlexiaJM 8y ago|0 comments
5. 5.▲
   by AlexiaJM 8y ago

   It's from same GAN though, C(x_r) and C(x_f) come from the same neural network. It's how realistic real data is compared to fake data (and vice-versa) as determined by C (discriminator without activation function).
6. 6.▲
   by AlexiaJM 8y ago

   "Relative probability" is only correct for standard GAN, not for other GANs which don't estimate a probability. My first name was "Critic's difference" since it's literally C(x_r)-C(x_f) so its the differe
7. 7.▲
   by AlexiaJM 8y ago

   Paper is here: https://arxiv.org/abs/1807.00734 . Feel free to ask me questions!
8. 8.▲
   The relativistic discriminator: a key element missing from standard GAN(ajolicoeur.wordpress.com)

   3 points by AlexiaJM 8y ago|1 comments
9. 9.▲
   Understanding GANs through a statistical divergence perspective(alexiajm.github.io)

   2 points by AlexiaJM 9y ago|0 comments