Engineering topics

Begoli, Edmon

Publications and source records attributed to Begoli, Edmon.

3 records

DOE OSTI2024

A Compound Data Poisoning Technique with Significant Adversarial Effects on Transformer-based Sentiment Classification Tasks

Transformer-based models have demonstrated much success in various natural language processing tasks. However, they are often vulnerable to adversarial attacks, such as data poisoning, which can intentionally fool the model into generating incorrect results. In this article, we present a novel, compound variant of a data poisoning attack on a transformer-based model that maximizes the poisoning effect while minimizing the scope of poisoning. Here we do so by combining the established data poisoning technique (label flipping) with a novel adversarial artifact selection and insertion technique aimed at minimizing detectability and the scope of the poisoning footprint. We find that by using a combination of these two techniques, we achieve a state-of-the-art attack success rate of approximately 90% while poisoning only 0.5% of the original training set, thus minimizing the scope and detectability of the poisoning action. These findings have the potential to advance the development of better data poisoning detection methods.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2024

Semantic Stealth: Crafting Covert Adversarial Patches for Sentiment Classifiers Using Large Language Models

Deep learning models have been shown to be vulnerable to adversarial attacks, in which perturbations to their inputs cause the model to produce incorrect predictions. As opposed to adversarial attacks in computer vision, where small changes introduced to pixel values can drastically alter a model's output while remaining imperceptible to humans, text-based attacks are difficult to conceal due to the discrete nature of tokens. Consequently, unconstrained gradient-based attacks often produce adversarial examples that lack semantic meaning, rendering them detectable through visual inspection or perplexity filters. In contrast to methods that rely on gradient-based optimization in the embedding space, we propose an approach that leverages a Large Language Model's ability to generate grammatically correct and semantically meaningful text to craft adversarial patches that seamlessly blend in with the original input text. These patches can be used to alter the behavior of a target model, such as a text classifier. Since our approach does not rely on gradient backpropagation, it only requires access to the target model's confidence scores, making it a grey-box attack. We demonstrate the feasibility of our approach using open-source LLMs, including Intel's Neural Chat, Llama2, and Mistral-Instruct, to generate adversarial patches capable of altering the predictions of a distilBERT model fine-tuned on the IMDB reviews dataset for sentiment classification.

Roa Carvajal, Maria↗

DOE OSTI2024

Diversity and scale: Genetic architecture of 2068 traits in the VA Million Veteran Program

One of the justifiable criticisms of human genetic studies is the underrepresentation of participants from diverse populations. Lack of inclusion must be addressed at-scale to identify causal disease factors and understand the genetic causes of health disparities. We present genome-wide associations for 2068 traits from 635,969 participants in the Department of Veterans Affairs Million Veteran Program, a longitudinal study of diverse United States Veterans. Systematic analysis revealed 13,672 genomic risk loci; 1608 were only significant after including non-European populations. Fine-mapping identified causal variants at 6318 signals across 613 traits. One-third (n = 2069) were identified in participants from non-European populations. This reveals a broadly similar genetic architecture across populations, highlights genetic insights gained from underrepresented groups, and presents an extensive atlas of genetic associations.

59 BASIC BIOLOGICAL SCIENCES↗