Analog Quantum Variational Embedding Classifier

Quantum machine learning has the potential to provide powerful algorithms for artificial intelligence. The pursuit of quantum advantage in is an active area research. For current noisy, intermediate-scale (NISQ) computers, various quantum-classical hybrid have been proposed. One such previously proposed algorithm a gate-based variational embedding classifier, which composed classical neural network and parameterized circuit. We propose classifier based on analog computer, where control signals vary continuously time. In our algorithm, data transformed into parameters time-varying Hamiltonian computer by linear transformation. nonlinearity needed nonlinear classification problem purely provided through dependence final state Hamiltonian. performed numerical simulations that demonstrate effectiveness performing binary multi-class linearly inseparable datasets as concentric circles MNIST digits. Our can reach accuracy comparable with best classifiers. find performance be increased increasing number qubits until saturates fluctuates. Moreover, optimization scales qubits. increase training when size increases therefore not fast network. presents possibility using annealers solving practical machine-learning problems, it could also useful explore learning.