On the generalizability of artificial neural networks in spin models

The applicability of artificial neural networks (ANNs) is typically limited to the models they are trained with and little is known about their generalizability, which is a pressing issue in the practical application of trained ANNs to unseen problems. Here, by using the task of identifying phase transitions in spin models, we establish a systematic generalizability such that simple ANNs trained with the two-dimensional ferromagnetic Ising model can be applied to the ferromagnetic qq-state Potts model in different dimensions for q \geq 2q≥2. The same scheme can be applied to the highly nontrivial antiferromagnetic qq-state Potts model. We demonstrate that similar results can be obtained by reducing the exponentially large state space spanned by the training data to one that comprises only three representative configurations artificially constructed through symmetry considerations. We expect our findings to simplify and accelerate the development of machine learning-assisted tasks in spin-model related disciplines in physics and materials science.

• Publication year

  2020

• Venue

  SciPost Physics Core

• Publication date

  2020-06-26

• Fields of study

  Physics, Computer Science

• Identifiers
  DOI 10.21468/SciPostPhysCore.5.2.032arXiv 2006.15021
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• High Bias

  2023cited by this paper
• Finding the deconfinement temperature in lattice Yang-Mills theories from outside the scaling window with machine learning

  2021cited by this paper
• Exploring neural network training strategies to determine phase transitions in frustrated magnetic models

  2020cited by this paper
• A Mapping of Distinct Phase Transitions to a Neural Network

  2020cited by this paper
• Machine learning and the physical sciences

  2019cited by this paper
• Generalization properties of neural network approximations to frustrated magnet ground states

  2019cited by this paper
• A high-bias, low-variance introduction to Machine Learning for physicists

  2018cited by this paper
• 92 Years of the Ising Model: A High Resolution Monte Carlo Study

  2018cited by this paper
• Smallest neural network to learn the Ising criticality.

  2018cited by this paper
• Machine learning phases of matter

  2016cited by this paper
• Learning phase transitions by confusion

  2016cited by this paper
• TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems

  2016cited by this paper
• Machine learning: Trends, perspectives, and prospects

  2015cited by this paper
• autoScale.py - A program for automatic finite-size scaling analyses: A user's guide

  2009cited by this paper
• Phase Transition Properties of 3D Potts Models

  2008influential reference
• Normalized entropy density of the 3D 3-state Potts model

  2007influential reference
• Spontaneous Magnetization in the Two-Dimensional Ising Model

  2003cited by this paper
• Mersenne twister: a 623-dimensionally equidistributed uniform pseudo-random number generator

  1998cited by this paper
• Antiferromagnetic Potts Models on the Square Lattice: A High-Precision Monte Carlo Study

  1998cited by this paper
• Logarithmic Corrections and Finite-Size Scaling in the Two-Dimensional 4-State Potts Model

  1996cited by this paper
• Three-state antiferromagnetic Potts models: A Monte Carlo study.

  1990cited by this paper
• Collective Monte Carlo updating for spin systems.

  1989cited by this paper
• CORRIGENDUM: Logarithmic corrections to finite-size scaling in strips

  1986cited by this paper
• The Potts model

  1982influential reference
• Exactly solved models in statistical mechanics

  1982cited by this paper
• Critical antiferromagnetic square-lattice Potts model

  1982cited by this paper
• “Model”

  1981cited by this paper
• Magnetism and local molecular field.

  1971cited by this paper
• Some generalized order-disorder transformations

  1952cited by this paper
• Crystal statistics. I. A two-dimensional model with an order-disorder transition

  1944cited by this paper

• Curvature-driven shifts of the Potts transition on spherical Fibonacci graphs: a graph-convolutional transfer-learning study

  2026cites this paper
• Discovering quasiorder parameters in the Potts model: A bridge between machine learning and critical phenomena

  2025influential citation
• A comprehensive neural network study of the non-equilibrium phase transition of the two-dimensional Ising model on the square lattice

  2024cites this paper
• Supervised and unsupervised learning of (1+1) -dimensional even-offspring branching annihilating random walks

  2023cites this paper
• Applications of Domain Adversarial Neural Network in phase transition of 3D Potts model

  2023cites this paper
• Learning the phase transitions of two-dimensional Potts model with a pre-trained one-dimensional neural network

  2023cites this paper
• Applications of Machine Learning to Lattice Quantum Field Theory

  2022cites this paper
• Study of phase transition of Potts model with Domain Adversarial Neural Network

  2022cites this paper
• Finding the deconfinement temperature in lattice Yang-Mills theories from outside the scaling window with machine learning

  2021cites this paper
• A Mapping of Distinct Phase Transitions to a Neural Network

  2020cites this paper
• Neural Network Statistical Mechanics

  2020cites this paper
• Mapping distinct phase transitions to a neural network.

  2020cites this paper
• Machine-learning physics from unphysics: Finding deconfinement temperature in lattice Yang-Mills theories from outside the scaling window

  2020cites this paper