Adversarial Machine Learning Phases of Matter

Abstract

We study the robustness of machine learning approaches to adversarial perturbations, with a focus on supervised learning scenarios. We find that typical phase classifiers based on deep neural networks are extremely vulnerable to adversarial perturbations: adding a tiny amount of carefully crafted noises into the original legitimate examples will cause the classifiers to make incorrect predictions at a notably high confidence level. Through the lens of activation maps, we find that some important underlying physical principles and symmetries remain to be adequately captured for classifiers with even near-perfect performance. This explains why adversarial perturbations exist for fooling these classifiers. In addition, we find that, after adversarial training, the classifiers will become more consistent with physical laws and consequently more robust to certain kinds of adversarial perturbations. Our results provide valuable guidance for both theoretical and experimental future studies on applying machine learning techniques to condensed matter physics.

arXiv comment5 pages main text plus 9 pages supplementary materials

References in corpus (62)

• Topological Insulators
• Topological insulators and superconductors
• Explaining and Harnessing Adversarial Examples
• Intriguing properties of neural networks
• Ensemble Adversarial Training: Attacks and Defenses
• Solving the Quantum Many-Body Problem with Artificial Neural Networks
• One pixel attack for fooling deep neural networks
• Machine learning phases of matter
• Wild Patterns: Ten Years After the Rise of Adversarial Machine Learning
• Transferability in Machine Learning: from Phenomena to Black-Box Attacks using Adversarial Samples
• Many-body quantum state tomography with neural networks
• Learning phase transitions by confusion
• Object Detectors Emerge in Deep Scene CNNs
• Discovering Phase Transitions with Unsupervised Learning
• Adversarially Robust Generalization Requires More Data
• Unsupervised learning of phase transitions: from principal component analysis to variational autoencoders
• Adversarial Attacks and Defences: A Survey
• Quantum Entanglement in Neural Network States
• Technical Report on the CleverHans v2.1.0 Adversarial Examples Library
• Machine learning quantum phases of matter beyond the fermion sign problem
• Discovering Phases, Phase Transitions and Crossovers through Unsupervised Machine Learning: A critical examination
• Efficient Representation of Quantum Many-body States with Deep Neural Networks
• Restricted-Boltzmann-Machine Learning for Solving Strongly Correlated Quantum Systems
• Machine Learning Topological States
• Identifying topological order through unsupervised machine learning
• Quantum Loop Topography for Machine Learning
• Machine Learning Topological Invariants with Neural Networks
• Machine Learning Phases of Strongly Correlated Fermions
• Identifying Quantum Phase Transitions using Artificial Neural Networks on Experimental Data
• Machine Learning in Electronic Quantum Matter Imaging Experiments
• Unsupervised machine learning and band topology
• Machine learning meets quantum physics
• Adversarial vulnerability for any classifier
• Deep Learning the Quantum Phase Transitions in Random Two-Dimensional Electron Systems
• Identifying Quantum Phase Transitions with Adversarial Neural Networks
• Machine Learning ${\mathbb{Z}}_2$ Quantum Spin Liquids with Quasi-particle Statistics
• Classifying Snapshots of the Doped Hubbard Model with Machine Learning
• Unsupervised Manifold Clustering of Topological Phononics
• Topological quantum phase transitions retrieved through unsupervised machine learning
• Unveiling phase transitions with machine learning
• Machine Learning Many-Body Localization: Search for the Elusive Nonergodic Metal
• Vulnerability of quantum classification to adversarial perturbations
• Parameter diagnostics of phases and phase transition learning by neural networks
• Machine Learning Spatial Geometry from Entanglement Features
• Machine Learning Bell Nonlocality in Quantum Many-body Systems
• Deep Learning the Quantum Phase Transitions in Random Electron Systems: Applications to Three Dimensions
• Machine Learning Topological Phases with a Solid-state Quantum Simulator
• Noncommutative geometry for three-dimensional topological insulators
• Unsupervised identification of topological order using predictive models
• Extrapolating quantum observables with machine learning: Inferring multiple phase transitions from properties of a single phase
• Unsupervised machine learning of quantum phase transitions using diffusion maps
• Probe of Three-Dimensional Chiral Topological Insulators in an Optical Lattice
• Unsupervised Learning of Non-Hermitian Topological Phases
• Drawing Phase Diagrams of Random Quantum Systems by Deep Learning the Wave Functions
• The critical temperature of the 2D-Ising model through Deep Learning Autoencoders
• Divergence of predictive model output as indication of phase transitions
• Adversarial Spheres
• Unsupervised learning using topological data augmentation
• Generalized No Free Lunch Theorem for Adversarial Robustness
• Interpretable and unsupervised phase classification
• Visualizing Quantum Phases And Identifying Quantum Phase Transitions By Nonlinear Dimensionality Reduction
• Featuring the topology with the unsupervised machine learning

Cited by in corpus (3)

• Towards quantum enhanced adversarial robustness in machine learning
• Universal adversarial perturbations for multiple classification tasks with quantum classifiers
• Artificial intelligence for representing and characterizing quantum systems