Research
- Published
- Authors
-
Gamaleldin Elsayed and Michael Mozer
Unveiling the Influence of Adversarial Images on Human Perception
Discovering the intricate connection between artificial and human vision.
New research reveals how subtle alterations in digital images can impact human perception, mirroring the effects on computer vision systems.
Exploring the blurred lines between human and AI perception.
Understanding Adversarial Images
Delving into the deceptive realm of adversarial attacks on digital imagery and AI systems.
Exploring the subtle yet profound effects of adversarial attacks on both digital and physical objects.
Delving into the intriguing implications of adversarial examples on human judgment.
Human Response to Adversarial Examples
Unmasking the hidden influence of adversarial images on human perception.
Peering into the subtle perceptual biases induced by adversarial images.
Connecting the dots between human perception and machine vision in the realm of adversarial images.
Implications for AI Safety and Security
Addressing the critical implications of human susceptibility to adversarial perturbations in AI systems.
Proposing strategies to enhance the robustness of computer vision models based on human perceptual alignment.
Emphasizing the imperative need for cognitive science research in shaping safer AI technologies.