Recent Breakthroughs in Quantum AI

Artificial intelligence (AI) is a technological breakthrough that has changed the way we live our daily lives. It has brought us technology we often take for granted—such as smart home devices—to tackling large world problems like climate change. Like many technologies, however, classical AI techniques are reaching their limits in terms of computing power. In the never-ending search for bigger, better, faster—enter Quantum AI.

Quantum AI uses quantum computing to improve computational tasks within AI and other related fields, such as machine learning (ML) and natural language processing (NLP). While seeming to be an ideal solution for the existing issues with classical AI, there have been some concern over “barren plateaus,” which occurs when optimization problems turn flat resulting in no clear path to a solution.

Recent research from Los Alamos National Laboratory (LANL), Absence of Barren Plateaus in Quantum Convolutional Neural Networks, published in Physical Review X, offered good news regarding Quantum AI and the risk of a barren plateau.

“The Los Alamos work shows how some quantum neural networks are, in fact, immune to barren plateaus,” says ScienceDaily on the LANL release. “The Los Alamos team developed a novel graphical approach for analyzing the scaling within a quantum neural network and proving its trainability.”

The key to LANL’s solution is the construction of the quantum neural network, with some being “immune” to barren plateaus explains Marco Cerezo, co-author of the LANL paper.

“We proved the absence of barren plateaus for a special type of quantum neural network,” says Cerezo. “Our work provides trainability guarantees for this architecture, meaning that one can generically train its parameters.”

The network proposed by LANL could be used by a variety of researchers attempting to solve the latest technological and scientific problems.

“With this guarantee in hand, researchers will now be able to sift through quantum-computer data about quantum systems and use that information for studying material properties or discovering new materials, among other applications,” said Patrick Coles, the paper’s other coauthor.

Additional breakthroughs in the field of Quantum AI were shared by IBM in July. The company announced they have developed a quantum kernel algorithm for a specific class of classification problems that is reproducibly faster than classical ML algorithms.

Published in Nature Physics, the paper written by Yunchao Liu from University of California, Berkeley and IBM research intern, alongside two IBM coauthors, describes the solution as “a rigorous and robust quantum speed-up.”

IBM’s algorithm uses an existing and proven machine learning model to trained on a quantum kernel method that can be efficiently solved in far less time than it would take with a classical method.

“Its quantum advantage comes from the fact that we can construct a family of datasets for which only quantum computers can recognize the intrinsic labeling patterns, while for classical computers the dataset looks like random noise,” shares IBM.

Quantum AI is still in its infancy in the world of AI and ML. As researchers spend more time developing the technology even more breakthroughs are bound to be discovered – potentially providing the answers to questions that haven’t yet been asked.

Capitol Tech offers bachelor’s, master’s, and PhD programs in cyber analytics and data science, including courses in machine learning, artificial intelligence, and natural language processing. Email admissions@captech.edu for more information.