Table 2 Mean and standard deviation of accuracy for six sub-sampled datasets of size 2000, equally distributed between two classes, with a batch size of 32.

From: Post-variational classical quantum transfer learning for binary classification

Test

Accuracy (%) ± SD

MLP

77.63 ± 0.028

ResNet50

80.60 ± 1.129

HQCNN

77.50 ± 0.015

CQTL Pennylane

78.90 ± 0.004

CQTL qiskit

79.90 ± 0.006

  1. For the HQCNN and MLP models, the final layer of the ResNet50 pre-trained model was modified to extract 8 features as input to the HQCNN and MLP models. The classical models considered are MLP and ResNet50. For the quantum models, HQCNN, CQTL in Pennylane, and CQTL in Qiskit were implemented with 8 qubits, using ResNet50 as the pre-trained model.
Back to article page