Generative and Sequential Deep Learning

Workshop Overview

This hands-on workshop dives deep into the rapidly evolving field of Generative and Sequential Deep Learning, focusing on the theory, applications, and practical implementation of models that can autonomously create data. Participants will explore how generative models such as Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), and Diffusion Models work, and how they transform industries, from art and media to healthcare and business analytics. It will also delve into Recurrent networks and their LSTM-empowered alternatives. This workshop is designed for data scientists, machine learning engineers, AI enthusiasts, and developers who have a basic understanding of deep learning and want to expand their expertise in generative models. Attendees should have experience with Python and be familiar with common machine learning frameworks like TensorFlow.

Learning Outcomes

• Gain a solid understanding of generative deep learning models' fundamental concepts and theories, including GANs, VAEs, and RNN and LSTM sequential models.
• Learn how to build and utilize these generative models effectively.
• Master the techniques for training generative models, covering loss functions, optimization strategies, and stability issues.
• Explore practical applications of generative models, such as image synthesis and text generation.
• Develop the skills to identify and troubleshoot common issues encountered during the training of generative models.
• Stay informed about the latest advancements and trends in generative deep learning research.

Duration 4 days

Day 1:

- Recurrent Neural Networks

- LSTM and GRU

Day 2:

-  Autoencoders.

Day 3:

-  Variational Autoencoders (VAE).

Day 4:

- Generative Adversarial Networks (GAN).

What will it be about?

- Comprehensive colored PPT documents.

- Cell state: forget, convey, ...

- Encoders and decoders

- Latent Space

- Auto Encoders Vs. Variational Auto Encoders

- Generators vs. Discriminators

- Objective Functions

- Mode Collapse

- Training approaches