Not only are membership inference attacks practical, cost-effective, and widely applicable in real-world scenarios, but recent advances in generative AI, particularly Large Language Models (LLMs), create novel challenges for membership privacy that…
Author: Brian Colwell
Alternative Investment Explorer
A Brief Taxonomy Of AI Membership Inference Attacks
In the below taxonomy, membership inference attacks are categorized by: target model, adversarial knowledge, attack approach, training method, and target domain. Target Model The target model category of this membership inference attack…
A Brief Taxonomy Of AI Membership Inference Defenses
In the below taxonomy, membership inference defenses are categorized as confidence masking, regularization, differential privacy, or knowledge distillation. Confidence Masking Confidence masking in machine learning is a technique where predictions with low…
The Bitter Reality Of AI Backdoor Attacks
In the rapidly evolving landscape of artificial intelligence, a silent threat lurks beneath the surface of seemingly trustworthy models: backdoor attacks. At its core, a backdoor attack is a method of compromising…
A Brief Introduction To AI Data Poisoning
As machine learning systems have become integrated into safety and security-sensitive applications at exponential speed, the responsible deployment of language models has increasingly presented complex challenges that extend beyond technical implementation: not…
A History Of Clean-Label AI Data Poisoning Backdoor Attacks
With significant advancements in stealth and effectiveness across diverse domains in just seven short years, the field of clean-label AI data poisoning has quickly evolved from the first major clean-label attack framework…
A History Of Label-Flipping AI Data Poisoning Attacks
Label-flipping is popular because of key advantages such as requiring not only minimal access to data, but minimal computational resources, as well. In addition to this attack’s low effort and low cost…
A Taxonomy Of Backdoor AI Data Poisoning Attacks
In this section, backdoor data poisoning attacks are divided into the following categories: Backdooring Pretrained Models Attacks that insert hidden malicious behaviors into models during the pretraining phase, before they are fine-tuned…
A Taxonomy Of AI Training Data Poisoning Attacks
In this brief taxonomy, training data poisoning attacks are divided into the following categories: Bilevel Optimization Poisoning Attacks These attacks frame the poisoning problem as a bilevel optimization where the attacker solves…
A Taxonomy Of AI Data Poisoning Defenses
We begin our taxonomy by dividing data poisoning defenses into three broad categories: Attack Identification Techniques, Attack Repair Techniques, and Attack Prevention Techniques, in which are then organized key research papers by defense type….