Membership inference attacks Courses and Certifications

Google TechTalks

Privacy Auditing of Large Language Models

Discover advanced techniques for detecting privacy leakage in LLMs through novel canary generation methods that outperform traditional membership inference attacks.

• YouTube
• 54 minutes
• On-Demand
• Free Video

IEEE

Membership Inference Attacks against Machine Learning Models

Explore membership inference attacks on machine learning models, focusing on techniques to determine if specific data was used in training, with implications for privacy and security in ML applications.

• YouTube
• 20 minutes
• On-Demand
• Free Video

IEEE

Membership Inference Attacks against Adversarially Robust Deep Learning Models

Explore membership inference attacks on adversarially robust deep learning models, examining the interplay between security defenses and privacy vulnerabilities in AI systems.

• YouTube
• 27 minutes
• On-Demand
• Free Video

Google TechTalks

Low Cost High Power Membership Inference Attacks

Explore robust, efficient membership inference attacks for machine learning privacy risk assessment. Learn to differentiate population and training data with minimal computational overhead.

• YouTube
• 41 minutes
• On-Demand
• Free Video

USENIX

Enhanced Label-Only Membership Inference Attacks with Fewer Queries

Discover DHAttack, a novel label-only membership inference attack that achieves superior performance with minimal queries, overcoming sample and direction diversity challenges.

• YouTube
• 16 minutes
• On-Demand
• Free Video

IEEE

Membership Inference Attacks from First Principles

Explore the fundamentals of membership inference attacks, their implications, and defense strategies in this comprehensive IEEE talk by leading security researchers.

• YouTube
• 22 minutes
• On-Demand
• Conference Talk

IEEE

Comprehensive Privacy Analysis of Deep Learning

Análisis exhaustivo de la privacidad en el aprendizaje profundo, destacando ataques de inferencia de caja blanca y vulnerabilidades en modelos centralizados y federados.

• YouTube
• 17 minutes
• On-Demand
• Conference Talk

Simons Institute

Generative Adversarial Models for Privacy and Fairness

Explore generative adversarial models for enhancing privacy and fairness in machine learning, covering differential privacy, GANs, and real-world applications in data publishing and encoding.

• YouTube
• 38 minutes
• On-Demand
• Free Video

Conf42

A Beginner's Guide to Adversarial Machine Learning

Explore adversarial machine learning attacks, their impact on AI systems, and effective mitigation strategies. Gain insights into poisoning, inference, extraction, and evasion attacks with real-world examples.

• YouTube
• 39 minutes
• On-Demand
• Free Video

USENIX

Through the Lens of LLMs - Unveiling Differential Privacy Challenges

Explores LLMs as potential attackers in differential privacy, focusing on Google's Topics API vulnerabilities. Reveals increased re-identification risks and challenges privacy claims using practical simulations and LLM-driven analysis.

• YouTube
• 10 minutes
• On-Demand
• Free Video

Databricks

Beyond the Privacy-Utility Tradeoff - Differential Privacy in Tabular Data Synthesis

Explore advanced techniques for generating differentially private synthetic data that maintains high utility while protecting sensitive information in AI applications.

• YouTube
• 19 minutes
• On-Demand
• Free Video

USENIX

Towards Lifecycle Unlearning Commitment Management - Measuring Sample-level Unlearning Completeness

Discover a novel framework for measuring machine unlearning effectiveness that addresses privacy concerns by quantifying sample-level completeness without costly retraining.

• YouTube
• 13 minutes
• On-Demand
• Free Video

USENIX

Defending Against Data Reconstruction Attacks in Federated Learning: An Information Theory Approach

Explore advanced defense strategies against data reconstruction attacks in federated learning through information theory, focusing on privacy preservation and efficient training methods.

• YouTube
• 11 minutes
• On-Demand
• Free Video

Google TechTalks

Dataset Inference Methods for Large Language Models - Detecting Training Data Usage

Explore techniques for detecting if large language models were trained on specific datasets, focusing on copyright implications and statistical testing methods for dataset inference.

• YouTube
• 44 minutes
• On-Demand
• Free Video

USC Information Sciences Institute

The Surprising Effectiveness of Membership Inference with Simple N-Gram Coverage

Discover how N-Gram Coverage Attack enables membership inference on black-box language models, outperforming existing methods while requiring only text outputs, with implications for privacy and copyright protection.

• YouTube
• 1 hour 3 minutes
• On-Demand
• Free Video