Explainable and Reliable AI (2025-Fall)

Course Instructor

Prof. Seong Tae Kim

Time & Location

Tue. 13:30-14:45, Thu. 13:30-14:45 @ College of Electronics Information, B05

Course Description

This course introduces the principles and techniques of Explainable Artificial Intelligence (XAI), emphasizing the importance of trust and transparency in modern AI systems. Students will explore model interpretability, explanation methods, uncertainty analysis, and evaluation approaches to understand how AI systems make decisions and how their reliability can be assessed.

Objectives

  • TUnderstand the necessity and significance of Explainable Artificial Intelligence (XAI).
  • Learn and apply the latest research trends and techniques in XAI.
  • Understand and utilize evaluation methods for assessing XAI models and technologies.

Schedule

  • Week 1: Introduction to Explainable AI
  • Week 2: Basics of Deep Learning
  • Week 3: Feature Visualization
  • Week 4: Network Dissection
  • Week 5: Neuron-Concept Association
  • Week 6: Feature Attribution: Gradient-based Approaches
  • Week 7: Feature Attribution: Class-activation Mappings
  • Week 8: Feature Attribution: Perturbation
  • Week 9: Evaluation
  • Week 10: Project Proposal
  • Week 11: Uncertainty Analysis
  • Week 12: Semantic Explanation
  • Week 13: Interactive AI & Applications
  • Week 14: Final Exam
  • Week 15: Final Presentation

Evaluation

The final grade will be determined according to the following percentage breakdown.

  • Final-Exam: 50%
  • Assigment: 20%
  • Presentation: 20%
  • Attendance: 10%

Teaching Assistant

Youngseob Won