Stack Exchange Network

Stack Exchange network consists of 174 Q&A communities including Stack Overflow, the largest, most trusted online community for developers to learn, share their knowledge, and build their careers.

Visit Stack Exchange

Questions tagged [gan]

GAN refers to Generative Adversarial Networks. Such networks is made of two networks that compete against each other. The first one generates new samples and the second one discriminates between generated samples and true samples.

0
votes
0answers
8 views

Can a Conditional GAN be used for a regression task, where the condition can take any continuous value?

I came across conditional GAN, where the classes can be provided as input to the GAN. I would like to know, if the same network can be used for classes which can take any continuous value.ie, the ...
0
votes
0answers
17 views

Dataset for GANs (logo generation)

I'm begining work on a new project that involves GANs. So far what I've learnt from some publications (e.g. this) is that these models require literally tonnes of images, e.g. 80K. The problem I'm ...
0
votes
0answers
10 views

Understanding distributions being concentrated on low dimensional manifold (as to GAN convergence)

I am working through some papers concerning convergence of Generative Adversarial Networks (GANs) as in Arjovsky (2017) and I have trouble understanding how to imagine a distribution being ...
1
vote
0answers
42 views

How to train the generator in a recurrent GAN (Keras)

I am trying to train a Recurrent GAN that is meant to generate geospatial movement data (sequences of 3-tuples of latitude, longitude and time). You may simply consider it a sequences of vectors with ...
1
vote
0answers
17 views

how to implement infoGAN's loss function in Keras' functional API

I have been trying to make an infoGAN based on this, but it is in pure tensorflow and I can't really figure out how I would implement the Q_loss in Keras (preferably the functional API). here is ...
0
votes
1answer
23 views

Evaluating performance of Generative Adverserial Network?

What is the best way to evaluate performance of Generative Adverserial Network (GAN)? Perhaps measuring the distance between two distributions or maybe something else?
1
vote
0answers
11 views

Confused about transpose convolution and tensor shapes in tensorflow GAN tuturial

https://github.com/tensorflow/tensorflow/blob/r1.11/tensorflow/contrib/eager/python/examples/generative_examples/dcgan.ipynb class Generator(tf.keras.Model): def init(self): super(Generator, ...
0
votes
2answers
54 views

Value Function of Generative Adversarial Network

My question concerns the notation used in the Value function of a GAN Does "x ~ p_data(x)" mean $$ E_{x\sim p_{data}}[log(D(x)]= \sum_{x} logD(x)(x)p_{data}(x)$$ ?? Many thanks,
2
votes
1answer
27 views

How to calculate $\phi_{i,j}$ in VGG19 network?

In the paper Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network by Christian Ledig et al., the distance between images (used in the loss function) is calculated from ...
2
votes
1answer
31 views

Can I use a GAN to increase my Dataset used for Image detection?

I am currently working on a machine learning project where I use the YOLO Algorithm to detect an object inside of a picture or video. The problem I face is that the specific image set (side-scan sonar)...
1
vote
1answer
20 views

EGAN Paper With Confusing Notation

I am reading a research paper that tries to fix some issues with GAN, but for one of the equations, the paper does not fully explain where it comes from and why it works. Although the overall ...
1
vote
0answers
25 views

Architecture Advice for training a GAN

I'm trying to create a model that generates worlds for a game. The game is 2-dimensional and small worlds sizes. I've sampled about 15,000 worlds that are about 200x200 blocks. Each block has its own ...
3
votes
0answers
17 views

Fully connected layer output explodes, but weights, gradients, and inputs all have sane values

I'm trying to train a GAN, and the architecture includes a fully connected layer before the output activation function. In my case, by the second training iteration this layer's output always explodes....
0
votes
0answers
29 views

GAN isn't stable

I built GAN. When I checked generator network and it is very unstable. This is my code for generator: ...
1
vote
0answers
188 views

Is it possible for a DCGAN to do regression? What are some examples of this?

I'm currently a student doing some machine learning projects, and I want to use generative adversarial networks to train some data to discern for example, how old someone is. The intended output is a ...
1
vote
0answers
138 views

Why use GAN in NLG?

I am interested in the GAN recently. There are many papers that recently applied GAN to NLG. I do not know much about NLP or NLG, but I wonder why I use GAN for NLG. For better quality? Or for many ...
2
votes
1answer
207 views

Gumbel Softmax vs Vanilla Softmax for GAN training

When training a GAN for text generation, i have seen many people feeding the gumbel-softmax from the generator output and feed into the discriminator. This is to bypass the problem of having to sample ...
1
vote
1answer
468 views

How to collect all variables as a list in tensorflow grouped as a function

I am trying to reproduce the cGAN network architecture introduced on the recent paper deep video portrait(2018, Standford) I have defined Generator as ...
0
votes
1answer
74 views

GAN's for data augmentation [closed]

I am working to augment my data using Generative Adversarial Networks, I have used Deep Convolutional GAN's for this purpose but they are not learning the right data distribution, so please suggest me ...
1
vote
0answers
34 views

Random Training set for GAN's [closed]

I have studies the gans in depth and some of its type like cycle, pix2pix, cgans. Now I want to generate random images from random distribution from generator. So I am creating a dataset with no ...
0
votes
1answer
270 views

GAN for inpainting an image

Is it possible to train a GAN model to inpaint an image taken from a specific setting(e.g. an office, woods, beach ...) after we have cropped out people of it? For example, I used this repo's ...
0
votes
0answers
14 views

How to decide the scales of value function estimates and reward function estimates in deep reinforcement learning?

I am basically working with an Inverse Reinforcement Algorithm called Generative Adversarial Imitation Learning. It has a generator working as an RL agent and Discriminator who can output a reward ...
2
votes
0answers
22 views

GAN generator output minibatch classes order

Let's take MNIST as example. I'm interested which classes(digits) images and in which order there are in generator output. I think it always generates [0, 1, 2, 3, 4, 5, 6, 7, 8 ,9] images is this ...
1
vote
1answer
46 views

Can GAN generate very happy facial expression

I have some dataset ${(x1,y1), (x2,y2)...(xn,yn)}$, where, $x$ is the picture of a facial expression,while $y$ is the fraction corresponding to their degree of the happiness (happy laugh: $y$ close to ...
3
votes
1answer
640 views

How can both generator and discriminator losses decrease?

In this paper there is a plot of how the loss of gans looks through epochs. Figure 2: These are of course averaged losses. How can both the discriminator loss and generator loss decrease? ...
3
votes
1answer
1k views

GAN vs DCGAN difference

I am trying to understand the key difference between GAN and DCGAN. I know that DCGAN uses a convolutional network. But: What data is better to push into GAN and what data fits better to DCGAN? ...
1
vote
1answer
76 views

GAN to generate a custom image does not work

I have been training a GAN in the cloud for some time now. I use Google's free credit. My laptop with a CPU doesn't seem to be up to the task. The image I want to generate is this. Even though the ...
4
votes
0answers
76 views

Can a GAN-like architecture be used for maximizing the value of a regression predictor?

I can't seem to convince myself why a GAN model similar to regGAN couldn't be modified to maximize a regression predictor (see the image below). By changing the loss function to the difference ...
1
vote
1answer
554 views

Could someone explain to me how back-prop is done for the generator in a GAN?

I'm not very familiar with neural networks, however, I though I understood the concept of back propagation as starting from the error in the output layer. Say, we have 3 neurons in the output layer ...
3
votes
1answer
70 views

GANs and grayscale imagery colorization

I am currently studying colorization of grayscale satellite imagery as part of my Master's internship. After looking for various machine learning techniques, I quickly decided to go for deep learning, ...
0
votes
0answers
67 views

How can I train Generative Adversarial Inverse Reinforcement Learning(GAIL) by feeding encoded state representations in the GAN architecture ?

In GAIL there is an inside step where we need to train a GAN. When training GAN we input expert's [State, Action] pair and Agent's predicted [State, Action] pair into the Discriminator. Let's say my ...
1
vote
1answer
64 views

GAN - why doesn't the generator nullify the noise input?

In GAN architecture, during training, what keeps the generator's output dependant on the input noise? Why don't the weights of the noise input become zero (plus a bias)? I would expect the generator ...
1
vote
1answer
269 views

WGAN is too slow what are some ways to tweak for speed ups?

I have implemented a vanilla GAN which gave good results very fast but it had a lot of mode collapse issue, because of this I learned about WGAN which suppose to fix this, in fact they claim they have ...
0
votes
0answers
60 views

How to better visualizing Earth Mover Distance in relation with generative adversarial networks?

After reading the original EMD paper from 1998 I am having a hard time trying to visualize the connections between their dirt pile example to generative adversarial networks. Everything is kind of ...
0
votes
0answers
109 views

training gan with iris data

I trained gan with iris data using pytorch. I expected that generator easily train iris data. However, result is so bad. Does anyone can give me advise about this problem...? The codes below is ...
0
votes
0answers
94 views

Is there any applications in Generative Adversarial Inverse Reinforcement Learning in real world problems?

I think IRL can be super awesome in understanding behavioral patterns, reason out activities etc. But when it comes to applying to real-world problems still IRL suffers dues to unknown dynamics and ...
5
votes
1answer
64 views

how to generate automatically images meshing up different shapes with a deep learning software?

My pursuite is to generate something like a grottesque(a kind of painting producing human-animals and plants hybrids). I need to do something like this paints in order to create an art exhibition. I ...
0
votes
1answer
37 views

Image to image translation from image to sketch using GANs, do I need to annotated the face at all?

So far I have read 2 papers on GANs and it seems like they are unsupervised networks that only uses supervision for the discriminator. If I want to translate an image from a picture to line art all I ...
1
vote
0answers
175 views

Generator loss not decreasing- text to image synthesis

I am implementing Scott Reed's paper on Generative Adversarial Text to Image Synthesis.(https://arxiv.org/pdf/1605.05396.pdf) The dataset I am using is a simple one, consisting of images of circles, ...
1
vote
1answer
282 views

GAN discriminator converging to one output

I am running a GAN on the MNIST dataset. As the GAN continues to train, the quality of the generator seems to get worse, and it even seems as if it is starting to converge to just one value. Epoch 7:...
1
vote
1answer
78 views

feature selection such that features that explain target but are not correlated with confounds are picked

Please suggest a feature selection technique which selects features such that they explain target well but are not correlated with given know confounds. This paper(section 2.3) suggests using a GAN ...
0
votes
0answers
302 views

One sided label smoothing in GANs

How does one-sided label smoothing make the discriminator more robust by reducing the confidence in correct class?
0
votes
1answer
226 views

Multimodal distribution and GANs [closed]

What is intuition behind multimodal distribution? and How does GANs generate samples from it?
0
votes
1answer
45 views

What is Teacher Helping technique?

I read this paper, but I am having trouble understanding what Teacher Helping technique (page 3) is in context of RNN. Can someone explain to me what it is? Please assume I don't have much ...
0
votes
0answers
42 views

DCGAN celebA collapse

I am trying to generate attributed faces using DCGAN. Therefore I changed the code of the original implementation to use celebA instead of mnist. After 15 iterations I am getting the following Here ...
0
votes
0answers
36 views

Generative adversarial networks for multiple distribution noise removal

I am working on a project where I need to denoise images, and my dataset is composed of a big chunk of pairs ...
0
votes
0answers
155 views

is it a normal behavior of WGAN-GP to have this cost values?

I'm training WGAN-GP on some large images where they are not normalized. based on their research papers i never saw this cost values as what Im encountering right now. using this implementation WGAN-...
1
vote
2answers
106 views

How to use deep learning to add local (e.g. repairing) transformations to images?

I want to train a neural network that removes scratches from pictures. I chose a GAN architecture with a generator (G) and a discriminator (D) and two sets of images scratchy and non-scratchy, with ...
0
votes
1answer
642 views

Using Generative Adversarial Network to generate Single Image

Most generative adversarial networks learn the distribution of the dataset and then generate a sample of about 10's to 100's of images with similar distribution . I wanted to know if there has ...
0
votes
0answers
78 views

Generative Adverserial Networks - SELU instead of RELU

I recently read a paper on Hight-Resolution Deep Convolutional Generative Adverserial Networks, or HDCGANs for short. And in the paper they have been using SELU instead of RELU. I gone through it ...