2024 Summer Seminar #

Introduction #

Welcome to the 2024 Summer Seminar on AI: Optimization, Theory & Responsibility. This session will delve into various aspects of artificial intelligence, focusing on optimization techniques, theoretical foundations, and responsible AI practices. Our goal is to foster a collaborative and enriching learning environment where participants can explore state-of-the-art research and practical applications in these domains.

Reference Material #

Participants are encouraged to refer to the following resource for foundational knowledge and deeper insights: Understanding Deep Learning.

Schedule #

• Dates: Every Monday, from July 22th on
• Location: Room 341, School of Business
• Online Tencent Meeting: 907-2153-6929

Format #

The Seminar will primarily be conducted offline, with potential online channels via Tencent Meeting or Bilibili. Each session will feature presentations by selected participants, with each sub-topic assigned to 1-3 individuals (to be confirmed two weeks in advance).

Content Outline #

Optimization (Week 1, July 22) #

• Gradient-based optimization: Basic Knowledge
• Bayesian optimization: Tutorial

AI Theory (Weeks 2-3, July 29, August 5) #

• Gradient flow: Gradient Flow
• Neural tangent kernel: NTK

Responsible AI (Weeks 4-6, August 5, 19 and Sept 2) #

• Bias and fairness: Bias and Fairness
• Explainability I: Local Post-hoc Explanations
• Bayesian Neural Networks and Uncertainty: Introduction to Bayesian Neural Networks

We look forward to your active participation and insightful contributions throughout this discussion group.

You can get #

• Knowledge about fundamental and advanced methodologies in deep learning.

• A proof of your study experience: this website - Every contribution you make to this seminar will be recorded

  Hope everyone to deeply involved in this seminar, you can

  • Serve as a keynote speaker
  • Contribute some blogs and publish on the website

• Group Study

  • Work together to sovle for some problems - research experience

  • High-efficent learning pace & Study atmosphere

More #

• There are no explicit grades in this seminar, the only and foremost objective is to learn together.
• If you are audience, free free to come or not. (update the topics and main contents in advance).
• If you are keynote speaker, there are requirements for your representation (discuss later)