Despite achieving over 95% accuracy on ImageNet, popular models like ResNet, VGG, and EfficientNet can be easily misled by carefully crafted adversarial examples using FGSM, revealing deep learning’s inherent vulnerability and prompting the need for robust defense strategies.
This talk explores how large AI models become overconfident, leading to bias and hallucinations, examines adversarial examples in vision and language, explains why data and algorithms cause these issues, and shows how reinforcement learning can teach models to admit uncertainty and align with human values.
This article presents a comprehensive overview of adversarial example research applied to captcha systems, covering the definition and history of adversarial attacks, geometric‑aware generation frameworks, FGSM‑based attack variants, experimental results, trade‑offs between image quality and attack strength, and future work such as AdvGAN integration.
This article explains how neural networks work, walks through the step‑by‑step training process of a convolutional model, showcases vivid visualizations of each layer, and demonstrates how tiny adversarial perturbations can dramatically alter predictions, highlighting the importance of AI security.
AI brings convenience but also new security challenges; this article explains the two main sources of AI safety issues, details adversarial example techniques, showcases applications such as face‑recognition attacks and robust captcha designs, and highlights Alibaba’s research and the IJCAI‑19 AI adversarial competition.
The article explains how the black‑box nature of deep learning creates security risks such as adversarial attacks, describes real‑world examples in autonomous driving and medical imaging, and showcases JD Security's explainable AI system that demystifies model decisions to improve AI safety and industry adoption.
