Generative Adversarial Networks (GANs) are neural networks that are trained in an adversarial manner to generate data mimicking some distribution.
Questions tagged [gan]
208 questions
44
votes
3 answers
Were generative adversarial networks introduced by Jürgen Schmidhuber?
I read on https://en.wikipedia.org/wiki/Generative_adversarial_networks :
[Generative adversarial networks] were introduced by Ian Goodfellow et al in 2014.
but Jurgen Schmidhuber claims to have performed similar work earlier in that direction…
Franck Dernoncourt
- 42,093
- 30
- 155
- 271
21
votes
6 answers
Can a GAN be used for data augmentation?
Can a generative adversarial network (GAN) be used for data augmentation (i.e. to generate synthetic examples that are added to a dataset)? Would it have any impact on the performance of a model trained on the augmented dataset?
ErroriSalvo
- 273
- 1
- 3
- 12
17
votes
1 answer
Notation: What does the tilde below of the expectation mean?
I am reading about variational auto encoders, and there is the below loss function:
$$l_i(\Theta,\phi) = - {\mathbb{E}}_{z\sim q} \left[\log p_\phi(x_i|z)\right] + KL(q_{\phi}(z_i|x)||p(z))$$
What does the notation $z\sim q$ under $\mathbb{E}$ mean?…
Code Pope
- 781
- 6
- 17
15
votes
3 answers
What does a 'tractable' distribution mean?
For example, in generative adversarial network, we often hear that inference is easy because the conditional distribution of x given latent variable z is 'tractable'. Also, I read somewhere that Boltzmann machine and variational autoencoder is used…
sirius27
- 161
- 1
- 3
12
votes
1 answer
How to understand Generative Adversarial Networks Discriminative distribution?
So I am currently studying Generative Adversarial Network and I read the paper by Goodfellow a few times now Generative Adversarial Nets and a few other papers in this field (DCGAN, CycleGAN, pix2pix, and a few others).
But I have been struggling…
Kalle
- 225
- 1
- 5
11
votes
2 answers
What are the current methods to check for GAN overfitting?
In generative modeling, the goal is to find a way for a model to output samples of some distribution $p_X$ given a lot of samples $x_1, \ldots, x_n$. In particular, we want sampling from our model $G$ to satisfy
$G(z)$ is a new example
$G(z)$ looks…
Anon
- 567
- 3
- 16
9
votes
1 answer
GAN: why is too-good discriminator supposed to provide "small error"?
Reading here, I found:
If the discriminator wins by too big a margin, then the generator
can’t learn as the discriminator error is too small.
This is something I read somewhere else as well, but I can't really get it. If the discriminator has a…
rand
- 427
- 1
- 5
- 10
9
votes
5 answers
Why is the Optimal Discriminator $D^{*}_G(x) = \frac{p_\text{data}(x)}{p_\text{data}(x) + p_g(x)}$ in Generative Adversarial Networks?
Proposition 1,
The optimal discriminator is
$$
D^{*}_G(x) = \frac{p_\text{data}(x)}{p_\text{data}(x) + p_g(x)}
$$
At the proof, I couldn't understand about change of variables with integral.
Why the first line is changed to second line?!
$$
V(G,D)…
user3704652
- 91
- 5
8
votes
1 answer
Expected value notation in GAN loss
I am reading Goodfellow's original paper on GANs.
What I struggle to understand is his notation of the subscript in expected values.
$$
\mathbb{E}_{\boldsymbol{x} \sim p_{data}(\boldsymbol{x})}\ldots
$$
If I understand it correctly then…
pixelneo
- 118
- 8
8
votes
2 answers
What does log-likelihood mean in the context of generative models like GANs?
I understand the general notion of likelihood as "probability to generate the data given parameters" (like here). But people use (log-)likelihood as a measure of "goodness" of a generative model.
But, e.g., let's take a look at Generative…
sygi
- 183
- 1
- 4
7
votes
1 answer
Is the only difference between conditional generative models and discriminative models the complexity of the modeled distribution?
One common way to define discriminative models is that they model $P(Y|X)$, where $Y$ is the label, and $X$ is the observed variables. Conditional generative models do something quite similar, but the modeled distribution of $Y$ tends to be more…
shimao
- 22,706
- 2
- 42
- 81
7
votes
0 answers
Transfer Learning on generative adversarial networks (GANs)
Is it possible to apply transfer learning to GANs and if so, what are some examples of someone having tried this?
Alberto MQ
- 71
- 1
- 4
7
votes
0 answers
How is the loss calculated in a Wasserstein GAN?
I'm trying to implement a Wasserstein GAN according to this blog post:
https://myurasov.github.io/2017/09/24/wasserstein-gan-keras.html
And it has a wasserstein loss of:
def d_loss(y_true, y_pred):
return K.mean(y_true * y_pred)
My…
user135237
- 271
- 1
- 3
7
votes
2 answers
Importance of choice of latent distribution in GAN's
When working with GAN's, is there a clear theoretical difference of choosing my latent distribution to be a uniform vs a normal distribution?
My intuition says that if I learn from a $z\sim N(\mu,\sigma)$ latent variable that has a small $\sigma$,…
Mecasickle
- 292
- 1
- 12
6
votes
0 answers
Usage of dropout in convolutional GANs with batch norm?
In DCGAN, dropout is not used in either generator or discriminator.
When using batch norm, are the benefits of dropout generally so marginal that is is not used?
If it is used, in what circumstances? Both discriminator and generator?
Tom Hale
- 2,231
- 3
- 13
- 31