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.
236
questions
0
votes
0
answers
22
views
Generator loss not decreasing while training GAN
I’ve been attempting to create a basic GAN to generate images using this database of flowers (https://www.robots.ox.ac.uk/~vgg/data/flowers/102/).
I’ve spent a few days on this, and largely based my ...
0
votes
0
answers
10
views
Data generation with correlated columns
I am working on a finance project where i have to work on a model that predicts the short term impact on the bonds market (on the Spreads yield) when a big trade happens based on it's duration, ...
- 1
0
votes
0
answers
17
views
In the GAN objective function, why do we first do we first find the D(x) that maximizes the objective function and then maximise wrt the generator?
The GAN objective function is optimised like this: argmin(argmax(L(G,D))) where the argmax finds the D (Discriminator) that maximises L(G,D). Why is it not the other way around, i.e. argmax(argmin(L(G,...
0
votes
0
answers
15
views
How to train a VQ Gan Based on Pretrained Language and Image Encoder?
I have trained the dual encoder model (separate language and vision encoders) using the ideas here.
However, this looks like an image look up model rather than an image generation model. What would ...
- 315
1
vote
1
answer
15
views
gnet loss is increasing, while my dnet loss is decreasing, what is wrong with the GAN?
I am developing an MIT open source GAN to generate synthetic structured health data, in the standard FHIR format. I have a running GAN, and it is using 200,000 Synthea Patient FHIR resources to train....
- 11
0
votes
0
answers
26
views
Help with running topology GAN (TopoGAN)
Link: https://github.com/TopoXLab/TopoGAN-ECCV2020
Sorry if this is the wrong place to ask, but I've been looking for help on how to work this out for a long time as I don't have much experience in ...
0
votes
0
answers
26
views
Generator loss keeps increasing while discriminator keeps decreasing
I am trying to build a GAN to generate LEGO images however my generator is not working at all. I have tried changing the learning rates but it caused the loss to go even more higher, sometimes into ...
1
vote
1
answer
120
views
timeGAN Model Retraining
I am using timeGAN from ydata-synthetic repo, and now question is about re-training the model.
Suppose we have trained a model, say synth1, based on a certain ...
- 157
0
votes
1
answer
40
views
Generate Synthetic Data Indicating Original data's Trend
I am using timeGAN from ydata-synthetic repo.
Given a trained model synth, we generate synthetic data by:
...
- 157
1
vote
1
answer
28
views
GAN Output Gradient Calculation
Loss function for discriminator, which needs to be maximized: -log(D(x)) + log(1-D(G(z))).
Loss function for generator, which needs to be maximized: log(D(G(z)))
What would the calculation of the loss ...
- 11
0
votes
1
answer
39
views
Improve image classification model with trained generator
Is it possible to improve an image classification model with a generator (trained class conditionally). (so this is same source/target distribution and same source/target task, so not domain ...
1
vote
1
answer
110
views
Questions on reproducibility of TimeGAN results
I am playing the timeGAN model, using the example code from ydata-synthetic repo.
To train the model, we used
synth.train(stock_data, train_steps=50000)
to ...
- 157
1
vote
0
answers
30
views
What is the l2-norm of a scalar
What is the meaning of the l2-norm when dealing with scalar values? I'm assuming it would be the same thing as taking the absolute value.
For context: I am trying to implement the clustering method ...
0
votes
0
answers
9
views
Help with trying to reimplement ADEC paper (clustering using AE and GAN). Encoder loss is not decreasing
I am trying to reimplement the ADEC paper (https://arxiv.org/abs/1909.11832) which mixes an autoencoder with a GAN network, but I am facing the issue that the encoder loss does not decrease.
I have ...
0
votes
0
answers
22
views
Resizing layer in adversarial GAN
A lot of questions have already been asked regarding resizing layer in sequential models but I couldn't find any that could solve my specific architecture.
I am creating an adversarial GAN. First an ...
1
vote
2
answers
90
views
Why is cycle consistency loss alone not sufficient to produce meaningful output?
Imagine an adaptation of CycleGAN, in which the discriminators were removed in lieu of using only cycle consistency loss. Well, it turns out that the original authors of Cycle Consistent Adversarial ...
0
votes
0
answers
11
views
Why does this cGAN model perform poorly when trained on a different machine?
I was sent a cGAN model python file from a friend + the dataset he used to train this model. For him, the model trained succesfully & was able to generate very accurate images. These were his ...
- 1
0
votes
0
answers
34
views
Interpret loss of WGAN
I have been working with WGAN-GP to produce text.
However, I have some issues interpreting the loss functions.
First of all, the loss seems awfully small. After the first 300 batches (which translate ...
1
vote
0
answers
32
views
In WGAN paper, why does clipping weights approximate Lipschitz function?
In Wasserstein GAN, it's explained that maximizing a certain formula over a set of K-Lipschitz functions approximates the 1-Wasserstein distance and they model the functions as NNs. That much I ...
- 11
0
votes
1
answer
108
views
I'm Looking for dataset with Name, Father's Name, Mother's name, Age, Gender, Ethnicity, ~Birthplace [closed]
I am new to Data Science and I don't know where to find datasets, I Google searched, Asked ChatGPT and friends, but couldn't find what I was looking for.
My project is to generate new names based on ...
- 1
0
votes
1
answer
261
views
Transpose Convolution Output Size
I have been learning GAN (Generative Adversarial Networks) lately and having a hard time understanding the output size for transpose convolution. Let's say I am using a Tensor of [1, 64, 1, 1] as an ...
- 37
0
votes
1
answer
201
views
How to build an image generation model for interior room design?
I want to build an image generator model of interior room design. This model should be able to generate an interior image of a living room/bedroom/hall/kitchen/bathroom. I have searched about it and ...
- 51
0
votes
0
answers
148
views
Getting error "Failed to find data adapter that can handle input" even after converting list to array
I am getting this error : ValueError: Failed to find data adapter that can handle input' I even changed the list to arrays but still the error keeps pooping up.
This is the code:
...
0
votes
0
answers
36
views
Help to find strategy of transfer learning with a conditional pix2pix model
so I trained a similar model to pix2pix GAN to generate output images. I condition my model on three different types:
there are 120 input images and 2 parameters (param_1, param_2). For each parameter,...
- 1
0
votes
1
answer
95
views
GAN Generator Backpropagation Gradient Shape Doesn't Match
In the TensorFlow example (https://www.tensorflow.org/tutorials/generative/dcgan#the_discriminator) the discriminator has a single output neuron (assume batch_size=1).
Then over in the training loop ...
- 3
0
votes
0
answers
15
views
Understanding the Pulse Extractor in this paper
I'm reading this paper about a neural vocoder for singing synthesis:
https://arxiv.org/pdf/2210.12740.pdf
I've implemented vocoders before, but this one discusses a novel pulse extractor:
T[i] is ...
- 101
0
votes
0
answers
35
views
Which combination would be more beneficial : Resnet-50 and SVM or Resnet-18 and GANs?
I'm trying to compare the two methods that were used for COVID-19 detection.
Given that both these methods have approximately the same accuracies
which method according to you would be more beneficial(...
0
votes
0
answers
23
views
GAN architectures for handling big size images
I am trying to generate new images in order to improve the number of samples in a dataset.
I have noticed that the majority of the GANs are trained with quite small images. For example, ...
- 695
2
votes
0
answers
94
views
Diffusion Model without Text Prompt?
I hope I understand diffusion models well enough that this isn't too dumb of a question but...
I'd like to (I think) augment a diffusion model to:
At training time accept CT scan images, then apply an ...
- 21
0
votes
1
answer
45
views
How to train encoder in BiGAN?
I have some difficulties training a BiGAN. In particular, the encoder seems not learning the map between the images x and the latent space z.
I have the following encoder:
...
- 1
0
votes
1
answer
20
views
How long is the generator pre-trained in SeqGAN?
I am reading up about SeqGAN and I am trying to understand the pretraining step better.
The authors claim they want to maximize the Maximum Likelihood Estimation on the dataset S by pretraining the ...
3
votes
1
answer
99
views
Why is it an advantage "that Markov chains are never needed" to obtain gradients?
In the original GAN (Generative Adversarial Network) paper, Generative adversarial networks by I. Goodfellow, J. Pouget-Abadie, M. Mirza et. al. they state an advantage of the GAN is "that Markov ...
- 33
1
vote
1
answer
118
views
Confusion on DCGAN generator project & reshape
I've been recently studying DCGAN.
I tried following implementation from the pytorch.org DCGAN tutorial and found that it (seemingly) lacks project & reshape layer, which is present in the diagram:...
- 11
1
vote
1
answer
78
views
Where is the VGG in Cartoongan?
When I read through the paper of Cartoongan [CartoonGAN: Generative Adversarial Networks for Photo Cartoonization], I was so confused about where is vgg located in the entire network.
Based on the ...
- 13
0
votes
2
answers
715
views
Why is my DCGAN not converging?
I'm training a tf DCGAN on the MVTec hazelnut dataset and I found some difficulties. The problem is that after a lot of epochs the generate does not produce some quality images.
My model is the ...
- 1
2
votes
0
answers
55
views
Minor error in Ian Goodfellow's GAN optimality proof
I've been thinking of a part of the proof of the optimality of GANs from the original paper, and I can't manage to solve what seems to be an error. The paper states that the maximum of the function $y ...
- 21
1
vote
0
answers
58
views
Why is my GAN diverging?
I am trying to train a GAN related to this project. At ~100k steps, the discriminator's fake loss suddenly went to 0, and real loss increased very high. (Can look at the plots below). What could be ...
- 119
0
votes
0
answers
20
views
Do we know why GAN-based data augmentation works?
Although I've seen many examples of GAN-generated synthetic data greatly improving the performance of models, I struggle to understand how this is possible.
Say we are training a classifier $h$ to ...
1
vote
0
answers
7
views
Constrain GAN so the subject is always the same
I want to build a face GAN, but I want to be able to control the 'camera angle' of the generated image, and I also want the subject (the generated face) to be the same every time.
If the above is ...
- 113
1
vote
0
answers
15
views
In which way GAN generator transforms the data(for transforming a noise to the data)?
I have the problem: I understood how GAN works in general, but I need information how it work detailed. The part I don't understand is how the random noise at input is transformed to data on the ...
- 11
1
vote
0
answers
22
views
Is it right to argue that a testing dataset is not needed when evaluating the performance of a GAN?
For my degree final project I have been working on a GAN to solve a certain image enhancement task. The problem I’m currently working on has an extremely limited number of datasets due to the physical ...
- 11
1
vote
0
answers
218
views
Super Resolution GAN with different input image size
I am trying to train a Super Resolution GAN. Following some materials on the web
I managed to train the first SRGAN model. To do that I took some high-resolution images (128x128 pixels) and downscale ...
0
votes
0
answers
47
views
WGAN with simplest data
I want to train WGAN (pyTorch) to generate this simple data.
Here below my generator and critic architechtures:
...
- 1
0
votes
0
answers
149
views
Predict sequence using seqGAN
I am trying to create a GAN model in which I am using this seq2seq as Generator and the following architecture as Discriminator:
...
- 53
0
votes
0
answers
389
views
Image normalization and reverse normalization: colors lost on image generation (GAN)
I'm working on a Gan.
Based on different papers, I use a Tanh activation function on the last layer of the generator. Which produces [-1,1] outputs.
To make this coherent, I use image normalization ...
- 3
2
votes
1
answer
347
views
RCNN to predict sequence of images (video frames)?
In the following work the authors apply a convolutional recurrent neural network (RNN) to predict the spatiotemporal evolution of microstructure represented by 2D image sequences. In particular, they ...
- 123
0
votes
1
answer
71
views
Gan paper: sampling the distribution
In the Gan paper it is said page 3 Figure 1:
"The lower horizontal line is
the domain from which z is sampled, in this case uniformly. The horizontal line above is part of the domain
of x. The ...
- 3
1
vote
1
answer
854
views
In my GAN model, the discriminator loss quickly descends to magnitudes of $10^{-4}$ while generator loss is at levels of 5+?
I am creating a Generative Adversarial Network (GAN) for generating artificial trading cards, but I am a complete novice in the field. The problem I'm consistently having is that my discriminator, ...
0
votes
1
answer
243
views
Connection between GANs and adversarial learning
Is there a connection between: "Adversarial Learning" (AL) and "Generative Adversarial Networks" (GANs)? Is it valid to say that GANs employ AL?
- 1
1
vote
1
answer
254
views
Random Noise in GANs
What does the "random input" from noise (random noise) mean in GANs?
Let's say I want to synthesize data such as name, age, income etc.. Does my generator know the original data sets (and ...
- 13