Search for: All records

Recent advances in Artificial Intelligence (AI) have brought society closer to the long-held dream of creating machines to help with both common and complex tasks and functions. From recommending movies to detecting disease in its earliest stages, AI has become an aspect of daily life many people accept without scrutiny. Despite its functionality and promise, AI has inherent security risks that users should understand and programmers must be trained to address. The ICE (integrity, confidentiality, and equity) cybersecurity labs developed by a team of cybersecurity researchers addresses these vulnerabilities to AI models through a series of hands-on, inquiry-based labs. Through experimenting with and manipulating data models, students can experience firsthand how adversarial samples and bias can degrade the integrity, confidentiality, and equity of deep learning neural networks, as well as implement security measures to mitigate these vulnerabilities. This article addresses the pedagogical approach underpinning the ICE labs, and discusses both sample activities and technological considerations for teachers who want to implement these labs with their students. 

Multi-label image recognition has been an indispensable fundamental component for many real computer vision applications. However, a severe threat of privacy leakage in multi-label image recognition has been overlooked by existing studies. To fill this gap, two privacy-preserving models, Privacy-Preserving Multi-label Graph Convolutional Networks (P2-ML-GCN) and Robust P2-ML-GCN (RP2-ML-GCN), are developed in this article, where differential privacy mechanism is implemented on the model’s outputs so as to defend black-box attack and avoid large aggregated noise simultaneously. In particular, a regularization term is exploited in the loss function of RP2-ML-GCN to increase the model prediction accuracy and robustness. After that, a proper differential privacy mechanism is designed with the intention of decreasing the bias of loss function in P2-ML-GCN and increasing prediction accuracy. Besides, we analyze that a bounded global sensitivity can mitigate excessive noise’s side effect and obtain a performance improvement for multi-label image recognition in our models. Theoretical proof shows that our two models can guarantee differential privacy for model’s outputs, weights and input features while preserving model robustness. Finally, comprehensive experiments are conducted to validate the advantages of our proposed models, including the implementation of differential privacy on model’s outputs, the incorporation of regularization term into loss function, and the adoption of bounded global sensitivity for multi-label image recognition.