Research into whether quantum computers, which process information using the unusual rules of quantum physics, could run machine learning tasks faster or better than ordinary computers. It is a research field rather than a working technology: existing quantum machines are small and error-prone, and whether they will ever give a practical advantage for learning is disputed.
Technical view
Approaches include quantum versions of linear algebra routines, variational circuits trained like neural networks, and quantum kernel methods. Proposed exponential speed-ups often assume classical data can be loaded into quantum states efficiently, which is itself a bottleneck, and several were matched by quantum-inspired classical algorithms, starting with Ewin Tang’s 2018 result on recommendation systems. Variational models face barren plateaus, where gradients vanish as circuits grow. The strongest case is for data that is itself quantum, such as outputs of physics experiments, and claims of practical advantage remain contested.
Related
Part of AI Terminology Index, section: Adjacent Fields.