quantum neural network

A quantum neural network is a machine-learning model that encodes data into qubits and processes it through parameterized quantum circuits, exploiting superposition and entanglement rather than classical weighted layers. In practice, most designs are hybrid: a classical computer tunes the circuit's parameters while the quantum device does the computations.

Yes — open-source frameworks like Xanadu's PennyLane, Google's TensorFlow Quantum (released in 2020), and Qiskit Machine Learning provide tools for building and training them. Models run on classical simulators or on real cloud quantum hardware such as IBM Quantum's processors.

Not demonstrably yet — no practical advantage over classical networks has been proven, and today's noisy intermediate-scale (NISQ) devices limit circuit depth and qubit counts. A known obstacle is the 'barren plateau' problem, identified by researchers including Jarrod McClean and colleagues in 2018, where training gradients vanish as circuits grow.

Main research applications include quantum chemistry and molecular simulation, classifying quantum-encoded data, and generative models like quantum GANs. The field is still largely experimental, with demonstrations running on hardware with tens to a few hundred physical qubits.