What is knowledge for if it is not shared? That is according to DigitalSreeni. Very well said. And your videos are very well made. I spent many early morning hours watching these. The best part is that you patiently explain some 'simple' details. What is simple to someone might not be so for someone else. -Alles Gute Herr Sreeni!
That sounds like a doctoral thesis :) In general, the approach doesn't matter as long as you are getting desired results. Also please keep 'Occam's razor' in mind all the time when picking an approach.
So, it sounds to me like random noise is the input to the generator while the discriminator contains the 'target' information, let's say an image. The generator network is trained using the discriminator data until the error loss is acceptable. Correct? If this is true, how is this any different from a standard ANN that is trained via supervision? Or am I getting something wrong? I'm trying to figure this out. Thank you. Good info.
The difference between a GAN and a standard supervised ANN is that a GAN does not require labeled training data. In a supervised ANN, each training example must have a corresponding label (e.g., "cat" or "dog" for an image classification task). In a GAN, the discriminator only needs to know whether the input sample is real or fake. This makes GANs well-suited for tasks where labeled data is scarce or expensive to obtain. Another difference between GANs and supervised ANNs is that GANs can be used to generate new data. For example, a GAN could be trained to generate new images of cats, even if the training data only contains images of dogs. This is because the generator is trained to produce realistic outputs, even if it has never seen those outputs before.
@@DigitalSreeni Thanks for the response, I will have to delve into this a little deeper because some of it is not completely clear; as an example, does the training data go into the discriminator side or the generator side......I know, it may sound obvious to you but I come from a strictly ANN background for implementation in mobile robot platforms, most of which the outputs don't require explicit labeling...the outputs are responses to the input patterns and the error signals are derived from external sensors. And lastly, if the generator doesn't compare it's output to the discriminator, then how does the network know when an image is correct? Feel free to contact me for further clarification...I have so many other questions, as well as questions about my own networks I've built with a special architecture that eliminates the need for backprop. Thank you again.
@@DigitalSreeni Thank You. Please see if you have time to make a video on capsule networks. These are very hard to understand for now and I am sure you'll make it easy for us!
can we say if loss is getting low then fake images is not generated and if loss is getting higher then fake images are generated ???when we have given noise data and image file to gans????/
can you make a video with using GAN to detect text not image (let say as ex: attack text & not attack text for site), where discriminator contain 2 layer?
I haven't heard of denoising images using GAN. Besides, GANs take way too much time to train so it may not be a practical solution for denoising images, unless someone already trains and provides a model we can work with.
Would it make sense to use an VAE as a generator and then train the discriminator based in the input- vs output data of the VAE? And I wonder if I could use the trained discriminator for anomaly detection. The thing is, I have acoustic data of a running machine that has never failed (and it should not, it is a giant 100 kil-tons steel wheel rotating at high-speed) and I want to model an early-warning-system. It seems like the discriminator would be a tool that can be used for this, since the data overall is fairly homogeneous.
@@DigitalSreeni thank you very much sir sir pls help me accessing apeer account i am a research scholar perusing research on image processing i can not create the apeer platform pls do help in using apeer
Can one or you apply a GAN on Time Series data? Would that be possible? Or combine it with SLTM ? And if possible would you be able to give as en example of it? For me combining different algos is hard. BTW, I really live your videos- they are gems! People just had not discovered them. I think your way of explaining is succinct, to the point, unburdened with noise and efficiently clear. Thank you very much! Cheers!
one of the best channels for Deep Learning in Images. Thank you Sir for these wonderful tutorials
No doubt in that, he is so humble despite of being so good,
What is knowledge for if it is not shared? That is according to DigitalSreeni. Very well said. And your videos are very well made. I spent many early morning hours watching these. The best part is that you patiently explain some 'simple' details. What is simple to someone might not be so for someone else.
-Alles Gute Herr Sreeni!
I watched at least 10 videos on GAN, this one cleared my mind the best what is happening in GAN how it actually works..
I am glad it helped.
A great video sir! Thank you soo much for the crystal clear explanation!
Great video, very practical. Keep sending more!
Nice and crystal clear explanation. keep continuing sir
Keep watching
Hi. Can you compare classical upsampling based high resolution image generation DNNs with SR-GAN? When and why we should prefer GANs?
That sounds like a doctoral thesis :)
In general, the approach doesn't matter as long as you are getting desired results. Also please keep 'Occam's razor' in mind all the time when picking an approach.
@@DigitalSreeni You said that always try to keep the system as simple as possible :)
So, it sounds to me like random noise is the input to the generator while the discriminator contains the 'target' information, let's say an image. The generator network is trained using the discriminator data until the error loss is acceptable. Correct? If this is true, how is this any different from a standard ANN that is trained via supervision? Or am I getting something wrong? I'm trying to figure this out. Thank you. Good info.
The difference between a GAN and a standard supervised ANN is that a GAN does not require labeled training data. In a supervised ANN, each training example must have a corresponding label (e.g., "cat" or "dog" for an image classification task). In a GAN, the discriminator only needs to know whether the input sample is real or fake. This makes GANs well-suited for tasks where labeled data is scarce or expensive to obtain.
Another difference between GANs and supervised ANNs is that GANs can be used to generate new data. For example, a GAN could be trained to generate new images of cats, even if the training data only contains images of dogs. This is because the generator is trained to produce realistic outputs, even if it has never seen those outputs before.
@@DigitalSreeni Thanks for the response, I will have to delve into this a little deeper because some of it is not completely clear; as an example, does the training data go into the discriminator side or the generator side......I know, it may sound obvious to you but I come from a strictly ANN background for implementation in mobile robot platforms, most of which the outputs don't require explicit labeling...the outputs are responses to the input patterns and the error signals are derived from external sensors. And lastly, if the generator doesn't compare it's output to the discriminator, then how does the network know when an image is correct? Feel free to contact me for further clarification...I have so many other questions, as well as questions about my own networks I've built with a special architecture that eliminates the need for backprop. Thank you again.
As always, very clear explanation, thanks!
Cool introduction to GANs ;)
After watching your videos I feel confident enough to create something amazing! thank you
Wonderful. I am sure you will create something amazing as coding is easy, you are limited by your creativity :)
@@DigitalSreeni Thank You. Please see if you have time to make a video on capsule networks. These are very hard to understand for now and I am sure you'll make it easy for us!
@@DigitalSreeni Yes Please make GAN for Video data with label ... like giving text find video
Will GAN be helpful in repairing broken letters in images after pre processing them for OCR ?
I think yes.....did you try this ?
can we say if loss is getting low then fake images is not generated and if loss is getting higher then fake images are generated ???when we have given noise data and image file to gans????/
can you make a video with using GAN to detect text not image (let say as ex: attack text & not attack text for site), where discriminator contain 2 layer?
Thank you so much sir. You teach much better than my professor.
Incredible video, great breakdown❤
Hi Dear sir
Is there is Any practicle project on GAN,s in your video list?with coding?
Other than video 126 where I showed mnist I do not have any other videos on this topic.
Sir, my GPU is NVIDIA GeForce RTX 2060 and I have 32GB RAM. Is it enough to work with GANs? Please reply Sir.
Thanks for this great demonstration
Sir am trying to locate the forged part of an image which deep learning architecture you advice me to work on
Can GAN be good for Data Augmentation for EEG ?
thank u so much sir, can you do a video on denoising ct images using generative adversarial networks
I haven't heard of denoising images using GAN. Besides, GANs take way too much time to train so it may not be a practical solution for denoising images, unless someone already trains and provides a model we can work with.
@@DigitalSreeni thank's a lot
Sir, could you please upload a video on how to code a DCGAN using vgg-19 for image colorization.
Great work! Love this video!
Thank you so much!
@@DigitalSreeni Are you a teaching researcher currently?
No.
You are amazing! You should win the Noble Prize for these educational series.
Can you please do Convnext or semantic segmentation using gan
Would it make sense to use an VAE as a generator and then train the discriminator based in the input- vs output data of the VAE? And I wonder if I could use the trained discriminator for anomaly detection.
The thing is, I have acoustic data of a running machine that has never failed (and it should not, it is a giant 100 kil-tons steel wheel rotating at high-speed) and I want to model an early-warning-system. It seems like the discriminator would be a tool that can be used for this, since the data overall is fairly homogeneous.
Thanks, very clear explaination.
thanks very much for good teaching. i also watched variational autoencoder and it was perfect.
Great intro! Thank you!
Glad you like it!
🙏 Thanks for the help
Amazing tutorial. Thank you
You're very welcome!
Thanks for detail explanation
Thanks for sharing, that was so so much good. Thanks a lot of sir
Thank you for this video is very helpful in understading GANs. Can you please provide the slides for this?
you are amazing Sir
very help full video
thank you for this vedio sir it is very informative sir can u pls suggest me the latest methods for denoising medical images?
ua-cam.com/video/yO15IISXA1Y/v-deo.html
@@DigitalSreeni thank you very much sir
sir pls help me accessing apeer account i am a research scholar perusing research on image processing i can not create the apeer platform pls do help in using apeer
Can we use GAN for anomalies detection?
is can be but its too hard I have work on video to text using stack LSTM
Thank You Sir..!!
You are welcome.
thank you very much .
Thank you sir
Can one or you apply a GAN on Time Series data? Would that be possible? Or combine it with SLTM ? And if possible would you be able to give as en example of it? For me combining different algos is hard. BTW, I really live your videos- they are gems! People just had not discovered them. I think your way of explaining is succinct, to the point, unburdened with noise and efficiently clear. Thank you very much! Cheers!
Make some videos on imitation learning plz
Thank u
Welcome
I believe
sir I have question. Can you send me your email.thanks