Related Experiment Video
Updated: Aug 31, 2025
09:11
Revealing Neural Circuit Topography in Multi-Color
Published on: November 14, 2011
15.1K
Experimental demonstration of adversarial examples in learning topological phases
Huili Zhang1, Si Jiang1, Xin Wang1
1Center for Quantum Information, IIIS, Tsinghua University, Beijing, 100084, P. R. China.
Nature Communications
|August 25, 2022
Summary
Researchers demonstrated adversarial examples in machine learning for classifying topological phases. Experimental noise can trick AI classifiers, highlighting vulnerabilities in applying machine learning to condensed matter physics experiments.
Area of Science:
- Condensed matter physics
- Quantum information science
- Machine learning applications
Background:
- Classifying phases of matter and their transitions is a key challenge in condensed matter physics.
- Machine learning (ML) offers promising new approaches for phase identification.
- The reliability of ML in experimental settings requires thorough investigation.
Purpose of the Study:
- To experimentally demonstrate adversarial examples in ML-based topological phase classification.
- To investigate the impact of experimental noise on ML classifier reliability.
- To highlight the vulnerability of ML techniques in experimental physics.
Main Methods:
- Utilized a nitrogen-vacancy (NV) center platform for experimental demonstration.
- Simulated topological phases, specifically Hopf insulators.
- Implemented neural network-based classifiers and introduced adversarial perturbations through data manipulation (dropping input data).
Main Results:
- Experimental noises were shown to act as adversarial perturbations, especially with increased data loss.
- Successfully created adversarial examples that deceived the phase classifier with high confidence.
- The underlying topological properties of the simulated Hopf insulators remained unchanged despite successful deception.
Conclusions:
- Machine learning classifiers are vulnerable to adversarial examples in experimental settings.
- Experimental noise can be exploited to create deceptive inputs for ML models.
- This work underscores the need for robust ML techniques in experimental condensed matter physics and quantum information.
Related Concept Videos
02:31Phase Transitions
19.7K
Whether solid, liquid, or gas, a substance's state depends on the order and arrangement of its particles (atoms, molecules, or ions). Particles in the solid pack closely together, generally in a pattern. The particles vibrate about their fixed positions but do not move or squeeze past their neighbors. In liquids, although the particles are closely spaced, they are randomly arranged. The position of the particles are not fixed—that is, they are free to move past their neighbors to...
19.7K
01:34Woodward–Hoffmann Selection Rules and Microscopic Reversibility
3.3K
Electrocyclic reactions, cycloadditions, and sigmatropic rearrangements are concerted pericyclic reactions that proceed via a cyclic transition state. These reactions are stereospecific and regioselective. The stereochemistry of the products depends on the symmetry characteristics of the interacting orbitals and the reaction conditions. Accordingly, pericyclic reactions are classified as either symmetry-allowed or symmetry-forbidden. Woodward and Hoffmann presented the selection criteria for...
3.3K
02:39Phase Diagrams
43.0K
A phase diagram combines plots of pressure versus temperature for the liquid-gas, solid-liquid, and solid-gas phase-transition equilibria of a substance. These diagrams indicate the physical states that exist under specific conditions of pressure and temperature and also provide the pressure dependence of the phase-transition temperatures (melting points, sublimation points, boiling points). Regions or areas labeled solid, liquid, and gas represent single phases, while lines or curves represent...
43.0K
02:33Phase Transitions: Sublimation and Deposition
17.5K
Some solids can transition directly into the gaseous state, bypassing the liquid state, via a process known as sublimation. At room temperature and standard pressure, a piece of dry ice (solid CO2) sublimes, appearing to gradually disappear without ever forming any liquid. Snow and ice sublimate at temperatures below the melting point of water, a slow process that may be accelerated by winds and the reduced atmospheric pressures at high altitudes. When solid iodine is warmed, the solid sublimes...
17.5K
01:19Phase Diagram
6.0K
The phase of a given substance depends on the pressure and temperature. Thus, plots of pressure versus temperature showing the phase in each region provide considerable insights into the thermal properties of substances. Such plots are known as phase diagrams. For instance, in the phase diagram for water (Figure 1), the solid curve boundaries between the phases indicate phase transitions (i.e., temperatures and pressures at which the phases coexist).
6.0K
01:27Associative Learning
541
Associative learning is a fundamental concept in behavioral psychology, wherein a connection is established between two stimuli or events, leading to a learned response. This process is critical in understanding how behaviors are acquired and modified. Conditioning, the mechanism through which associations are formed, can be divided into two main types: classical conditioning and operant conditioning, each elucidating different aspects of associative learning.
Classical conditioning, also known...
Classical conditioning, also known...
541