Researchers from Northwestern University, Boston College, and MIT have developed a groundbreaking synaptic transistor inspired by the human brain. The device, introduced in a study titled "Moiré synaptic transistor with room-temperature neuromorphic functionality" published in Nature, exhibits capabilities beyond simple machine learning, showcasing its potential for higher-level thinking.
Unlike previous brain-like computing devices constrained to cryogenic temperatures, this new transistor remains stable at room temperatures. It operates at high speeds, consumes minimal energy, and retains stored information even without power, making it suitable for practical applications.
Mark C. Hersam of Northwestern University, a co-leader of the research, emphasizes the device's departure from the traditional digital computer architecture. In the brain, memory and information processing are integrated, resulting in significantly higher energy efficiency compared to digital computers.
Hersam explains, "Our synaptic transistor similarly achieves concurrent memory and information processing functionality to more faithfully mimic the brain."
Motivated by recent advancements in artificial intelligence (AI), the researchers sought to develop computers that emulate human brain operations. The current challenge with digital computing systems lies in their separate processing and storage units, leading to energy-intensive tasks. The study proposes an alternative to the prevailing silicon architecture, addressing the escalating power consumption associated with big data.
The researchers explored moiré patterns, a geometric design arising when two patterns layer atop each other. By combining bilayer graphene and hexagonal boron nitride and manipulating their relative positions, they achieved neuromorphic functionality at room temperature, offering a promising avenue for AI and machine learning.
Hersam explains the significance of this new design parameter: "With twist as a new design parameter, the number of permutations is vast."
The team trained the transistor to recognize similar patterns, demonstrating its associative memory—a higher-level form of cognition. The device successfully identified patterns even when presented with incomplete or imperfect input, showcasing its potential to handle complex real-world conditions.
Hersam envisions advancing AI technology toward higher-level thinking, addressing the limitations of current algorithms in handling intricate situations. The breakthrough in neuromorphic transistors opens new possibilities for energy-efficient, brain-inspired computing, with implications for diverse applications, from self-driving vehicles to complex data analysis.
