QML: Quantum Machine Learning Courses

QML -- quantum machine learning -- applies quantum computing to ML tasks and uses classical ML to optimize quantum algorithms. The field is still research-stage, but moving quickly. These are the best QML courses available today, ranked by rating.

Quantum gate geometric structure representing quantum machine learning circuit architecture

What is QML?

QML is the intersection of quantum computing and machine learning. Rather than running classical neural networks on quantum hardware (which does not work well), QML researchers build circuits that behave analogously to ML models: they have trainable parameters, accept input data, and can be optimized by minimizing a loss function.

The most practical QML models today are variational quantum circuits (VQCs), also called parameterized quantum circuits (PQCs). These run on NISQ devices -- noisy, near-term quantum hardware -- and are trained classically using gradient descent, with the quantum circuit evaluated at each step.

Variational quantum circuits

Parameterized circuits trained by gradient descent. The parameters (rotation angles) are updated classically; the circuit runs on quantum hardware or a simulator. The most common QML architecture today.

Quantum kernel methods

Quantum computers can compute certain kernel functions exponentially faster than classical hardware. Quantum kernels feed into support vector machines and other kernel-based classifiers -- one of the most theoretically grounded QML speedup claims.

Quantum neural networks

Layered VQCs designed to mimic classical neural networks. They suffer from barren plateaus -- vanishing gradients at scale -- which limits their current trainability. An active area of QML research.

QML frameworks: PennyLane vs Qiskit Machine Learning

Two frameworks dominate QML development. Which to use depends on your ML background and target hardware.

PennyLane
Qiskit Machine Learning
Developer
Xanadu
IBM
ML integration
PyTorch, TensorFlow, JAX
scikit-learn, PyTorch
Auto-differentiation
Yes (native)
Yes (via PyTorch)
Hardware access
IBM, AWS Braket, plugins
IBM Quantum
Best for
QML research, hybrid models
IBM ecosystem users

PennyLane is the de facto standard for QML research. If you're starting from scratch, learn PennyLane first. If you're already working with Qiskit for other reasons, Qiskit Machine Learning integrates cleanly.

QML courses

All courses with quantum machine learning content, sorted by rating.

QML frequently asked questions

What does QML stand for?
QML stands for quantum machine learning -- the application of quantum computing to machine learning tasks. The field explores whether quantum algorithms can provide speedups for ML workloads, and also uses classical ML techniques to optimize quantum algorithms. Most QML research today focuses on variational quantum circuits (VQCs) as trainable models.
Is QML the same as quantum machine learning?
Yes. QML is the widely used abbreviation for quantum machine learning. The two terms are used interchangeably in research papers, course titles, and industry discussions. You may also see "quantum ML" or "quantum AI" used loosely, though quantum AI has a broader meaning that includes quantum optimization and other non-ML applications.
What framework do most QML researchers use?
PennyLane, developed by Xanadu, is the dominant QML framework. It integrates directly with PyTorch, TensorFlow, and JAX, supports automatic differentiation of quantum circuits, and has a large library of built-in quantum ML layers and datasets. Qiskit Machine Learning is the main alternative for those already working in the IBM ecosystem.
Does QML currently outperform classical machine learning?
Not on any real-world task at practical scale. Current quantum hardware (NISQ devices) is too noisy and limited in qubit count to run the circuits needed for genuine quantum ML advantage. Theoretical speedups have been proven for specific problem classes, but these require quantum-accessible data and fault-tolerant hardware -- neither of which is widely available yet. QML is a research-stage field.
What math do I need for QML?
QML requires linear algebra (matrices, eigenvalues, inner products), probability and statistics, and calculus (for gradient-based optimization). You should also have working knowledge of classical ML: gradient descent, loss functions, and at minimum supervised learning fundamentals. Quantum mechanics background helps but is not strictly required to get started with PennyLane or Qiskit Machine Learning.