Adversarial AI: Poking Holes in Classification Algorithms

Overview

Machine learning models are prone to adversarial attacks, where inputs are manipulated to cause misclassification. While previous research has focused on techniques like Generative Adversarial Networks (GANs), there's limited exploration of GANs and Synthetic Minority Oversampling Technique (SMOTE) in text classification models. Our study addresses this gap by training various machine learning models and using GANs and SMOTE to generate additional data points aimed at attacking these models. Furthermore, we extend our investigation to face recognition models, training a CNN and subjecting it to adversarial attacks with perturbations. Our experiments reveal a significant vulnerability in classification models. Specifically, we observe a 13% decrease in accuracy for the top -performing text classification models post -attack, along with a 66% decrease in facial recognition accuracy. This h ighlights the susceptibility of these models to manipulation of input data. Adversarial attacks not only compromise the security but also undermine the reliability of machine learning systems. By showcasing the impact of adversarial attacks on both text classification and face recognition models, our study underscores the urgent need for robust defenses against such vulnerabilities. Addressing these vulnerabilities is crucial for ensuring the trustworthiness and effectiveness of machine learning applications across various domains.

Competition history

• JSHS 2024 Category not listed

Resources

• National JSHS abstract book ↗

Related projects

Closest projects by meaning, across every fair and year in the corpus.