Researchers at Washington University study complex problem-solving. They created NeuroSA, a tool aimed at solving hard problems.
NeuroSA tackles optimization problems, which involve finding the best solution among many. Examples include planning supply chains or designing new drugs. The tool uses a neuromorphic architecture that mimics the brain’s neurons and synapses.
The researchers have described the methods and results of this study in a paper published in Nature Communications.
NeuroSA also leverages principles of quantum mechanical behavior, specifically Fowler-Nordheim (FN) annealers. The FN annealers in NeuroSA take inspiration from quantum tunneling, a process where particles pass through barriers, to search for solutions efficiently.
“NeuroSA’s structure is neuromorphic, like our brain structure with neurons and synapses, but its search behavior is determined by the FN annealer," said research leader Shantanu Chakrabartty in a press release. "That critical bridge between neuro and quantum is what makes NeuroSA so powerful and what allows us to guarantee we’ll find a solution if given enough time.”
This combination makes NeuroSA powerful. It guarantees finding the best solution if given enough time. Unlike traditional methods, NeuroSA reliably discovers new solutions. The tool runs on SpiNNaker2, a neuromorphic computing platform. SpiNNaker2 is hardware that supports brain-like processing. Tests show NeuroSA works in real-world settings.
Applications and Future Potential
NeuroSA suits problems needing strategic shifts. It could optimize logistics in manufacturing or transportation. In drug discovery, it may find ideal molecular structures. Search times vary from days to weeks, depending on problem complexity. The tool’s design ensures better results than traditional computer models.
Chakrabartty sees NeuroSA accelerating solutions in supply chains and healthcare. The research, supported by the National Science Foundation and other groups, began at a workshop. Collaborators include SpiNNcloud Systems, which develops SpiNNaker2.
NeuroSA’s method transforms how we approach discovery problems. Discovery problems involve finding unknown solutions in machine learning. According to the reserachers, the tool offers a reliable way to design efficient systems and innovate in science. It marks a step toward solving real-world challenges with brain-like, quantum-inspired technology.