Noise-injected analog Ising machines enable ultrafast statistical sampling and machine learning

Ising machines are a promising non-von-Neumann computational concept for neural network training and combinatorial optimization. However, while various neural networks can be implemented with Ising machines, their inability to perform fast statistical sampling makes them inefficient for training neural networks compared to digital computers. Here, we introduce a universal concept to achieve ultrafast statistical sampling with analog Isingmachines by injecting noise. With an opto-electronic Ising machine, we experimentally demonstrate that this can be used for accurate sampling of Boltzmann distributions and for unsupervised training of neural networks, with equal accuracy as software-based training. Through simulations, we find that Ising machines can perform statistical sampling orders-of-magnitudes faster than software-based methods. This enables the use of Ising machines beyond combinatorial optimization and makes them into efficient tools for machine learning and other applications.

Automated summary

Ising machines are a promising computational concept for neural network training and combinatorial optimization. However, their inefficiency in fast statistical sampling hinders their performance compared to digital computers. In this study, we introduce a universal concept of using noise injection to achieve ultrafast statistical sampling with analog Ising machines, enabling accurate sampling of Boltzmann distributions and unsupervised training of neural networks. Through simulations, it is found that Ising machines can perform statistical sampling orders-of-magnitudes faster than software-based methods, making them efficient tools for machine learning and other applications.