Thursday, February 1, 2024 - 12:00 to 13:00
Introduction to Generative Adversarial Networks (GANs)
Presenter: Prithvika Babu
Date: Thursday, February 1, 2024
Time: 12:00 pm - 1:00 pm
Location: 4th Floor (Workshop space) at 300 Ouellette Avenue (School of Computer Science Advanced Computing Hub)
LATECOMERS WILL NOT BE ADMITTED once the presentation has begun.
Generative AI has taken the world by storm with the art, music, videos, etc. that it generates based on the inputs e.g., ‘draw a dog in a hat’ or ‘make this rock music classical’. This popular concept involves deep learning, and one of the popular deep learning models that are currently being used is the Generative Adversarial Networks (GAN). This model is based on the zero-sum game theory and its significance is to obtain the data distributed through unsupervised learning. To generate more realistic/actual data, it makes use of a ‘generator’ and ‘critic’ model working together, where the generator will generate the output and the critic will evaluate the generated outcome and provide feedback accordingly. This process is iterated until a final output is produced, which is then shown to the end-user. In this workshop, a brief introduction to generative AI and GAN, the methodology of GAN, and the existing problems of GANs are summarized, and the future work of GANs models which include the possible future applications in other media, as well as non-entertainment sectors.
Brief Introduction to Generative AI
Introduction to GAN
Explaining the methodology of GAN
Generator and Discriminator Functions
Statistical Approach – Probability Spaces
Explaining the limitations of GAN
Explaining the future works of GAN
In this workshop, we will be introduced to Generative AI, and one of the most common models used in Generative AI which is called GAN i.e., generative adversial network. The working of this model will be explained in detail, starting from the basic working to the statistics and mathematics involved in the functions. Then we will go over the limitations of the model and finish with the future works of GANS, current applications and possible future applications. The seminar will conclude with a QA session.