Description:

In this talk, we will explore Diffusion models, a revolutionary class of generative models that have captured the imagination of the creative community. They offer a unique approach to generating high-dimensional sequential data such as images, and with the help of diffusion models, we can produce a diverse variety of artistically realistic as well as non-photorealistic creations within the realm of generative art. During this talk, we will understand the basic principles behind diffusion models and delve into Stable Diffusion. Then, we will see how to personalize a Stable Diffusion model using textual inversion, with a demo of HuggingFace's DreamBooth.

Intended Audience:

Creative coders, artists, researchers, and enthusiasts of generative art and machine learning.

Outline of the talk:

• Introduction to Generative Models for AI Art - 5 minutes
• Diffusion Models and Stable Diffusion - 10 to 12 minutes
• (Python) Demo with DreamBooth - 10 minutes
• Links to additional resources, Key takeaways - 3 to 5 minutes

Prerequisites:

Following are the pre-requisites for the session:

• Fundamentals of Machine Learning
• Basics of Generative Models [desirable, not mandatory]
• Familiarity with HuggingFace & OpenAI (DALL.E)

Please note that while these pre-requisites provide a foundation for understanding the talk, the presenter will adjust the level of technicality and complexity to cater to the audience's knowledge and interests. Additionally, the talk may include brief explanations and examples to ensure that attendees can grasp the main concepts, regardless of their background in the field.

Speaker Info:

Nazia is a post-graduate student at IIIT Lucknow with a specialization in machine learning. She is passionate about human languages and AI art. She recently completed an internship with Fidelity Investments, a global fintech firm in the capacity of a Data Scientist Intern. Previously, she has been a Lead ML Engineer at Omdena, and a Research Intern at the University of Copenhagen, Denmark. She is the founder of the AIM Data Science Club at her campus, which is an initiative to create a collaborative environment for aspiring machine learning engineers and data scientists. As a Teaching Assistant for the Python programming lab at her institute, she regularly mentored students through their Python journey, helping them take their first steps into the world of technology. Nazia recently attended the Oxford ML Summer School where she found a newfound boost to her passion for AI Research. Her work was recently awarded first at Maitreyee, IBM Research's annual research showcase event. She also attended Association for Computational Linguistics (ACL) 2023 where two of her research papers were accepted.

Speaker Links:

1. https://www.linkedin.com/in/nazianafis/
2. https://medium.com/@nazianafis/
3. https://github.com/nazianafis/