“`html
Solving a 3×3 Rubik’s cube is straightforward, stated Shantanu Chakrabartty, the Clifford W. Murphy Professor and associate dean for research and graduate education in the McKelvey School of Engineering at Washington University in St. Louis. Simply learn and memorize the processes, then implement them to reach the solution. Computers excel at this type of procedural problem-solving. Now, Chakrabartty and his team have created a tool that transcends procedure to uncover innovative solutions to intricate optimization challenges in various fields such as logistics and drug development.
Chakrabartty and his collaborators unveiled NeuroSA, a neuromorphic problem-solving architecture inspired by human neurobiology that utilizes quantum mechanical behavior to identify optimal solutions — with certainty — and does so more consistently than leading-edge techniques. This collaborative effort across multiple universities, published on March 31 in Nature Communications, originated at the Telluride Neuromorphic and Cognition Engineering workshop and was spearheaded by Chakrabartty and lead author Zihao Chen, a graduate student in electrical and systems engineering at McKelvey Engineering.
“We are exploring methods to address problems more effectively than traditional computers designed around human learning,” Chakrabartty remarked. “NeuroSA aims to tackle the ‘discovery’ challenge, the most difficult issue in machine learning, focused on identifying new and unknown solutions.”
In optimization, annealing serves as a technique for assessing various potential solutions before ultimately selecting the optimal one. Fowler-Nordheim (FN) annealers harness quantum mechanical tunneling principles to efficiently seek that best solution, and they serve as the “secret ingredient” in NeuroSA, according to Chakrabartty.
“In optimization scenarios, strategy is critical when the system must adapt — for instance, when searching for the tallest building on campus, at what point do you venture into a different area?” Chakrabartty explained. “NeuroSA’s design is neuromorphic, mimicking the structure of our brain with neurons and synapses, while its search behavior is governed by the FN annealer. That essential connection between neuromorphic and quantum is what empowers NeuroSA and enables us to guarantee a solution if sufficient time is provided.”
This guarantee becomes particularly crucial when the timeframe for allowing NeuroSA to find an optimal solution could span from days to weeks, or even longer, based on the complexity of the problem. In the publication, Chakrabartty’s team, in partnership with the research group at SpiNNcloud Systems, has already demonstrated that NeuroSA can be integrated into the SpiNNaker2 neuromorphic computing platform, affirming its practical viability. Looking ahead, Chakrabartty envisions the tool being utilized to optimize logistics in supply chains, manufacturing processes, and transportation services or to reveal new drugs by examining optimal protein folding and molecular configurations.
Chen Z, Xiao Z, Akl M, Leugring J, Olajide O, Malik A, Dennler N, Harper C, Bose S, Gonzalez HA, Samaali M, Liu G, Eshraghian J, Pignari R, Urgese G, Andreou AG, Shankar S, Mayr C, Cauwenberghs G, Chakrabartty S. ON-OFF neuromorphic ISING machines employing Fowler-Nordheim annealers. Nature Communications, published online March 31, 2025. DOI: https://www.nature.com/articles/s41467-025-58231-5
This research received partial support from the U.S. National Science Foundation (ECCS:2332166 and FET:2208770), Germany’s Federal Ministry of Education and Research in the EIC Transition program under the “SpiNNode” project, the Horizon Europe project “PRIMI”, the Italian National Recovery and Resilience Plan (NRRP), the U.S. Department of Energy’s Office of Science, as well as the Advanced Manufacturing and Materials Technology Office of the Office of Energy Efficiency and Renewable Energy and SLAC National Laboratory.
SpiNNaker2 is a neuromorphic hardware accelerator platform developed by SpiNNcloud Systems, a commercial organization with which some co-authors have relations and financial interests. Washington University in St. Louis manages the patents and intellectual property rights associated with the Fowler-Nordheim based dynamic systems.
Originally published on the McKelvey Engineering website
The post A neuro-quantum leap in discovering optimal solutions appeared first on The Source.
“`