qubithubtrainable-kernel

Quantum kernel with learnable parameters optimized via kernel-target alignment (KTA). Interleaves data encoding (angle encoding) with trainable RZ/RY rotations and CNOT entanglement, then uses PennyLane autograd to maximize KTA — a differentiable proxy for classification performance. Demonstrates that trained kernels consistently outperform random baselines.

Run Cite

Framework: PennyLane
Qubits: 4
Depth: 22
Gate set: RY, RZ, CX
Licence: Apache-2.0
Version: v1
Updated: 1 day ago
Runs: None completed

circuit.py · 19 KB

Lineage

Not a fork.

No forks yet.

Suggested citation

@software{qubithub_trainablekernel_2026,
  author = {QubitHub Circuits},
  title = {Trainable Quantum Kernel},
  year = {2026},
  version = {v1},
  url = {https://qubithub.co/qubithub/trainable-kernel},
}

Generated from this circuit's metadata. QubitHub does not mint DOIs, so this is not a registered identifier — check it against your venue's requirements before publishing.

Code

Lineage

Suggested citation