Quantum ML
A complete quantum machine learning stack with three classifiers, six feature encoding strategies, and three quantum-aware optimizers. All algorithms follow the sklearn fit/predict/score pattern for familiar usage and integrate with PyTorch through a hybrid classical-quantum bridge.
QML Components
Variational Quantum Classifier
VQC with ZZ and Pauli feature maps, RealAmplitudes ansatz, and configurable layers. sklearn-compatible fit/predict/score API for binary and multi-class classification via one-vs-rest.
Quantum Support Vector Machine
QSVM computes quantum kernels K(x,y) = |⟨φ(x)|φ(y)⟩|² in exponential feature space. Backed by classical SVM with precomputed kernel matrices and automatic kernel caching.
Variational QNN
VariationalQNN with hardware-efficient and custom ansatz types. Supports both COBYLA (gradient-free) and SPSA optimization methods with validation tracking.
6 Feature Maps
ZZ Feature Map (entangling), Pauli Feature Map (multi-axis rotations), IQP (polynomial interactions), Amplitude Encoding, Angle Encoding, and Adaptive Feature Map (learnable parameters).
3 Quantum Optimizers
SPSA (gradient-free, 2 evaluations per iteration, noise-resistant), COBYLA (constrained optimization by linear approximation), and QuantumAdam (adaptive learning rates for quantum circuits).
Training Infrastructure
QMLTrainer with QuantumCrossEntropy, QuantumMSE, and FidelityLoss functions. Model save/load support, batch training, and convergence detection.