KAIST team makes quantum leap with new AI algorithmKAIST announced Tuesday that it has developed a quantum artificial intelligence (AI) algorithm that could boost the current state of artificial intelligence technology.
A team at KAIST led by June-koo Kevin Rhee, a professor of electrical engineering, worked with teams in Germany and South Africa to develop an artificial intelligence algorithm for nonlinear quantum computer machine learning.
The algorithm succeeded in a demonstration on the IBM Cloud service.
The newly developed algorithm needs fewer computation quantities compared to current technologies.
Quantum computers use subatomic particles called quantum bits, or quibits, as information. Because the quantum bits have the ability to be in multiple states simultaneously, they not only can handle large amounts of information, but also can solve complicated problems at an alarming speed.
This allows problems that have previously been unsolvable with existing computers to be solved.
However, quantum computers have been more geared toward solving linear equations, making it difficult to apply nonlinear machine learning with complicated data, which is crucial in artificial intelligence.
Rhee and his team developed a quantum algorithm that efficiently calculates the similarities within quantum data by combining quantum forking and quantum measuring technologies, allowing for supervised learning.
Kernel is the measure that quantifies similarities in data used in machine learning.
“The kernel-based quantum machine learning algorithm developed by the research team will replace existing supervised learning once the noisy intermediate-scale quantum computing era, which handles several 100s of qubits, expected to be commercialized in a few years, opens,” said Daniel K. Park, a KAIST professor on the research team.
BY LEE HO-JEONG [email@example.com]