Machine learning of high dimensional data on a noisy quantum processor

Abstract Quantum kernel methods show promise for accelerating data analysis by efficiently learning relationships between input points that have been encoded into an exponentially large Hilbert space. While this technique has used successfully in small-scale experiments on synthetic datasets, the practical challenges of scaling to circuits noisy hardware not thoroughly addressed. Here, we present our findings from experimentally implementing a quantum classifier real high-dimensional taken domain cosmology using Google’s universal processor, Sycamore. We construct circuit ansatz preserves magnitudes typically otherwise vanish due growing space, and implement error mitigation specific task computing kernels near-term hardware. Our experiment utilizes 17 qubits classify uncompressed 67 dimensional resulting classification accuracy test set is comparable noiseless simulation.