Table 8 The CQTL model was implemented using Qiskit and evaluated on three datasets.

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

Baseline model comparisons

Dataset

Accuracy (%)

Precision (%)

Recall (%)

F1-Score (%)

Runtime (s)

Ants and Bees

(Ep:30, Lr:0.004, Bs:4)

96.73

96.79

96.73

96.73

\(\approx 15\)

Cats and Dogs

(Ep:10, Lr:0.004, Bs:8)

98.00

97.90

98.00

98.00

\(\approx 17\)

Shirts and Pullover

(Ep:10, Lr:0.004, Bs:8)

95.40

95.60

95.30

95.40

\(\approx 15\)

  1. The first two datasets Ants vs. Bees and Cats and Dogs were selected based on optimal performance observed during the hyperparameter tuning phase for CQTL. To introduce greater dataset diversity and validate generalizability, the Shirts vs. Pullover dataset was also included. The specific hyperparameters used for each dataset are detailed in the corresponding experimental sections.