Quantum AI simulator using a hybrid CPU-FPGA approach
share
The quantum kernel method is one of the most important methods in quantum machine learning. However, the number of features used for quantum kernels has been limited to several tens of features. Here we use a block product state structure as a quantum feature map and demonstrate a field programmable gate arrays (FPGA) implementation. We show that our hybrid CPU-FPGA quantum kernel simulator is orders of magnitude faster than a conventional quantum computing simulator. This co-design of our quantum kernel and its efficient FPGA implementation enabled us to perform the largest numerical simulation of a gate-based quantum kernel in terms of input features, up to 780-dimensional features using 4000 samples. We apply our quantum kernel to image classification tasks using Fashion-MNIST dataset and show that our quantum kernel is comparable to Gaussian kernels with the optimized bandwidth. Our results have implications for developing quantum-inspired algorithms and designing quantum kernels.
READ FULL TEXT
