Neuromorphic devices and machine learning combine to make brain-like devices possible

As modern manufacturing increasingly relies on artificial intelligence (AI), automation, and real-time data processing, the need for faster and more energy-efficient computing systems has never been greater. A promising solution lies in neuromorphic computing—an emerging technology that mimics how the human brain processes information.

Unlike traditional computers, which process data in a linear and energy-intensive manner, neuromorphic devices are designed to work in parallel and adapt dynamically to new inputs. This brain-like approach allows them to handle complex tasks—such as recognizing patterns, analyzing images, and making decisions—more efficiently than conventional systems.

A new review published in the International Journal of Extreme Manufacturing offers a detailed look at recent advances in this field. Led by Prof. Zhong Lin Wang and Prof. Qijun Sun of the Beijing Institute of Nanoenergy and Nanosystems, together with Prof. Jeong Ho Cho of Yonsei University, the review outlines how researchers are combining machine learning algorithms with innovative hardware to build neuromorphic devices, and how these systems could be used in real-world applications in advanced artificial intelligence, hypermorphic computing, and interactive technologies.

The authors explain how machine learning models—such as support vector machine (SVM), artificial neural network (ANN), convolutional neural network (CNN), recurrent neural network (RNN), reservoir computing (RC)—are being embedded into physical devices that function like biological neurons and synapses. These neuromorphic chips can "learn" from data, adjust to changing environments, and process information in real time—capabilities that could significantly enhance smart manufacturing systems.

One of the key technical advances discussed in the review is the development of 3D neuromorphic device arrays. These dense networks of interconnected components resemble the brain's architecture and support massively parallel information processing, ideal for applications requiring high-speed, low-power computation. Such devices are already being tested in sensory systems, including artificial vision and touch, where they demonstrate the ability to process and respond to stimuli autonomously.

For the manufacturing industry, the implications are far-reaching. Neuromorphic systems could enable machines to better sense their environment, adapt to new tasks, and make decisions without relying on cloud computing or large amounts of external data. This would pave the way for more autonomous factories, intelligent robots, and real-time quality control systems, all operating with reduced energy consumption and greater efficiency.

However, challenges remain before neuromorphic devices can be widely deployed. As the pace of traditional chip development slows and the demand for AI processing grows, current systems must become more precise, reliable, and energy-efficient. The review points to promising materials innovations, such as replacing traditional insulating layers with solid-state electrolytes like ion gels, that could help improve performance and reduce power usage.

Looking ahead, the authors emphasize the need to further miniaturize neuromorphic components, integrate them into larger systems, and explore their use in brain-inspired computing platforms.

By more closely replicating biological information processing, such systems could enable new forms of artificial intelligence, more efficient computing for big data, and sophisticated human–machine interfaces. As research progresses, the line between biological and artificial cognition may continue to blur.

International Journal of Extreme Manufacturing (IJEM, IF: 21.3) is dedicated to publishing the best advanced manufacturing research with extreme dimensions to address both the fundamental scientific challenges and significant engineering needs.

• Maintain #1 in Engineering, Manufacturing for consecutive years
• Average time to First Decision after Peer Review: 34 days
• Open Access Publishing with APC Waivers

Visit our webpage, like us on Facebook, and follow us on Twitter and LinkedIn.