Machine learning generated configurations in presence of a conserved quantity: a cautionary tale

Abstract

We investigate the performance of machine learning algorithms trained exclusively with configurations obtained from importance sampling Monte Carlo simulations of the two-dimensional Ising model with conserved magnetization. For supervised machine learning, we use convolutional neural networks and find that the corresponding output not only allows to locate the phase transition point with high precision, it also displays a finite-size scaling characterized by an Ising critical exponent. For unsupervised learning, restricted Boltzmann machines (RBM) are trained to generate new configurations that are then used to compute various quantities. We find that RBM is incapable of recognizing the conserved quantity and generates configurations with magnetizations and energies forbidden in the original physical system. The RBM generated configurations result in energy density probability distributions with incorrect weights as well as in wrong spatial correlations. We show that shortcomings are also encountered when training RBM with configurations obtained from the non-conserved Ising model.

arXiv comment21 pages, 9 figures, submitted for publication to the Physical Review E

References in corpus (32)

• Solving the Quantum Many-Body Problem with Artificial Neural Networks
• Machine learning phases of matter
• Many-body quantum state tomography with neural networks
• Learning phase transitions by confusion
• Discovering Phase Transitions with Unsupervised Learning
• Unsupervised learning of phase transitions: from principal component analysis to variational autoencoders
• Predicting Crystal Structures with Data Mining of Quantum Calculations
• Machine learning quantum phases of matter beyond the fermion sign problem
• Discovering Phases, Phase Transitions and Crossovers through Unsupervised Machine Learning: A critical examination
• Restricted-Boltzmann-Machine Learning for Solving Strongly Correlated Quantum Systems
• Learning Thermodynamics with Boltzmann Machines
• Accelerate Monte Carlo Simulations with Restricted Boltzmann Machines
• Mutual Information, Neural Networks and the Renormalization Group
• Machine learning vortices at the Kosterlitz-Thouless transition
• Tomography and Generative Data Modeling via Quantum Boltzmann Training
• Detection of phase transition via convolutional neural network
• Neural Network Renormalization Group
• Kernel methods for interpretable machine learning of order parameters
• Unsupervised Learning of Frustrated Classical Spin Models I: Principle Component Analysis
• Machine learning of phase transitions in the percolation and XY models
• Restricted Boltzmann Machines for Quantum States with Nonabelian or Anyonic Symmetries
• Quantum-assisted Helmholtz machines: A quantum-classical deep learning framework for industrial datasets in near-term devices
• Self-Learning Monte Carlo Method in Fermion Systems
• Scale-invariant Feature Extraction of Neural Network and Renormalization Group Flow
• Machine-Learning Studies on Spin Models
• Smallest Neural Network to Learn the Ising Criticality
• Applications of neural networks to the studies of phase transitions of two-dimensional Potts models
• Machine Learning of Frustrated Classical Spin Models. II. Kernel Principal Component Analysis
• Learning the Ising Model with Generative Neural Networks
• Self-learning projective quantum Monte Carlo simulations guided by restricted Boltzmann machines
• Few-shot machine learning in the three-dimensional Ising model
• Identifying Product Order with Restricted Boltzmann Machines

Cited by in corpus (3)

• Machine learning phase transitions of the three-dimensional Ising universality class
• Learning a compass spin model with neural network quantum states
• Machine learning in physics: The pitfalls of poisoned training sets