uniqc.algorithms.core.training.qnn module

Quantum Neural Network (QNN) classifier with TorchQuantum backend.

Provides QNNClassifier — an nn.Module for binary classification using Hardware-Efficient Ansatz and TorchQuantum’s native autograd.

class uniqc.algorithms.core.training.qnn.QNNClassifier(n_qubits=4, n_features=2, depth=2)

Bases: Module

Quantum Neural Network for binary classification.

Uses angle encoding for input features and HEA for variational layer. Output is σ(<Z₀>) for binary classification.

Parameters:

• n_qubits – Number of qubits.

• n_features – Input feature dimension.

• depth – HEA depth.

forward(x)

Classify input batch.

Parameters:

x – Input tensor of shape (batch_size, n_features).

Returns:

Probability tensor of shape (batch_size,).