- Diese Veranstaltung hat bereits stattgefunden.
Team TensorFlow 18: GAN (Generative Adversarial Network)
August 8 @ 5:30 pm
We finished our discussion about object detection models. Our next topic are Generative Adversarial Network (GAN). They evolved from the fact that trained CNNs can be easily fooled by minimal perturbations in the input image which are imperceptible for humans.
We are still following the list at [The-9-Deep-Learning-Papers-You-Need-To-Know-About](https://adeshpande3.github.io/adeshpande3.github.io/The-9-Deep-Learning-Papers-You-Need-To-Know-About.html)
Please bring your notebook. It might be needed for code execution.
# Must Read
• Read the summary section about Generative Adversarial Networks [The-9-Deep-Learning-Papers-You-Need-To-Know-About](https://adeshpande3.github.io/adeshpande3.github.io/The-9-Deep-Learning-Papers-You-Need-To-Know-About.html)
• Read the paper by Ian Goodfellow from 2014: https://arxiv.org/pdf/1406.2661v1.pdf
• Read the Quora post from Yann LeCun: https://www.quora.com/What-are-some-recent-and-potentially-upcoming-breakthroughs-in-deep-learning
• Train the GAN network implemented in this repo (Vanilla GAN): https://github.com/wiseodd/generative-models
• Understand adversarial examples and the idea behind GANs
• Augment the GAN implementation with own ideas (e.g. exports to Tensorboard)
Location: Informatik-Gebäude des KIT (50.34), Raum -109
# Link to community repo:
# Take a look at the ML-KA homepage