Intrinsic Geometric Vulnerability of High-Dimensional Artificial Intelligence
arXiv:1811.03571
Abstract
The success of modern Artificial Intelligence (AI) technologies depends critically on the ability to learn non-linear functional dependencies from large, high dimensional data sets. Despite recent high-profile successes, empirical evidence indicates that the high predictive performance is often paired with low robustness, making AI systems potentially vulnerable to adversarial attacks. In this report, we provide a simple intuitive argument suggesting that high performance and vulnerability are intrinsically coupled, and largely dependent on the geometry of typical, high-dimensional data sets. Our work highlights a major potential pitfall of modern AI systems, and suggests practical research directions to ameliorate the problem.
References in corpus (8)
- Very Deep Convolutional Networks for Large-Scale Image Recognition
- The Loss Surfaces of Multilayer Networks
- Characterizing Adversarial Subspaces Using Local Intrinsic Dimensionality
- Estimating the intrinsic dimension of datasets by a minimal neighborhood information
- Adversarial Risk and the Dangers of Evaluating Against Weak Attacks
- Qualitatively characterizing neural network optimization problems
- Adversarial vulnerability for any classifier
- Local minima in training of neural networks