AIC Seminar "Generative AI Theory and Implementation" will be held on Thursday, August 1 and Friday, August 2, 2024.

■ Seminar Overview

We will lecture on generative models using deep learning (primarily image generation via diffusion models), combining theory and implementation, using "Deep Learning from Scratch: Generative Models (by Koki Saitoh)" as a reference book.

In recent years, diffusion models have been utilized in many image generation models such as Stable Diffusion, OpenAI's DALL-E 2, and Google's Imagen. The learning objective of this seminar is to literally become able to create image generation models from scratch, aiming for an understanding of the basic principles of generative models.

Let's learn to actually handle them through theoretical study and implementation.

【Date and Time】

8/1 (Thu), 8/2 (Fri) 13:00-15:00

Number of sessions: 4 in total (2 sessions per day x 2 days)

【Venue】

AIC Lounge, 2nd Floor, Kyoseikan Collaboration Complex, Hiyoshi Campus

【Format】

In-person

※There are no plans for recording or online streaming.

【Prerequisites】

You can attend without any problems if you have basic knowledge of probability and statistics, and experience with Python.

We recommend studying the content up to Step 3 of the reference material using any book or method.

Even if you have no experience with these, we will review all content as appropriate, although the pace is a bit fast, so you should be able to attend until the final session with sufficient study. We also plan to adjust the difficulty level based on surveys, so please feel free to participate.

【Target Audience】

Keio students, faculty, and staff

【Organizer, etc.】

AIC

【Pre-registration】

Required

※Keio students can apply here

Please see here for details regarding application contents, etc.