As computer vision models are being increasingly deployed in the real world, including
applications that require safety considerations such as self-driving cars, it is imperative that
these models are robust and secure even when subject to adversarial inputs.
This workshop will focus on recent research and future directions for security problems in
real-world machine learning and computer vision systems. We aim to bring together experts
from the computer vision, security, and robust learning communities in an attempt to highlight
recent work in this area as well as to clarify the foundations of secure machine learning. We
seek to come to a consensus on a rigorously framework to formulate adversarial machine
learning problems in computer vision, characterize the properties that ensure the security of
perceptual models, and evaluate the consequences under various adversarial models. Finally,
we hope to chart out important directions for future work and cross-community collaborations,
including computer vision, machine learning, security, and multimedia communities.
Topics:
- Theoretic understanding of adversarial machine learning
- (Related with vision based interpretable machine learning)
- Adversarial attacks against 3D computer vision tasks, e.g. 3D object detection for
autonomous driving
- Robust 3D deep learning models
- Real world attack models against current computer vision models
- Real world data distribution drift and its implications to model generalization and
robustness
- Repeatable experiments adding to the knowledge about adversarial examples on image,
video, audio, point cloud and other data
- Detection and defense mechanisms against adversarial examples
- (Related with robust computer vision systems)
- Novel challenges and discoveries in adversarial machine learning for computer vision
systems
- Vulnerabilities and potential solutions to adversarial machine learning in real-world
applications, such as autonomous driving

Papers should be within 5 pages excluding references and follow CVPR main conference paper format.