We are pleased to announce the upcoming AI So-Go-Chi Research Seminar, featuring a talk by Associate Professor Prarinya Siritanawan (Graduate School of Science and Technology, Shinshu University). The seminar will be held in a hybrid format on November 11 (Tuesday), from 13:00 to 14:00. We warmly welcome your participation and encourage you to share this information with colleagues who may be interested.

Date & Time: November 11, 13:00–14:00
Venue: Hybrid (On-site and Online)

• On-site: Room M153, Main Building, Research Institute of Electrical Communication (RIEC),
  Tohoku University
• Online (Zoom):
  • Meeting ID: 932 3469 8741
  • Passcode: 825120

Talk Information

Title: Toward Trustworthy Image Recognition: Understanding and Mitigating Adversarial Examples
Speaker: Assoc. Prof. Prarinya Siritanawan (Graduate School of Science and Technology, Shinshu University)

Abstract:
Adversarial examples are a form of attack that introduce subtle perturbations to images, causing recognition models to produce incorrect outputs. This talk examines the challenges these attacks pose to image recognition systems, with a particular focus on adversarial patches that can mislead models in real-world environments. We analyze the image features involved in such attacks and identify distinctive characteristics of adversarial patches that differ from common assumptions in previous studies. The talk also discusses mitigation approaches aimed at improving model robustness, contributing to the development of more secure and trustworthy image recognition systems.