Researchers at Northwestern University, Boston College, and the Massachusetts Institute of Technology have taken inspiration from the human brain to develop a new synaptic transistor capable of higher-order thinking, processing and storing information at the same time as the human brain. In the new experiment, the researchers demonstrated transistors' ability to classify data, go beyond simple machine learning tasks, and perform associative learning. The research results were published in the journal Nature on the 20th.
Although previous studies have used similar strategies to develop brain-inspired computing devices, these transistors can only operate at low temperatures. In contrast, the new equipment is stable at room temperature. It consumes very little energy when running at high speeds and retains stored information even when power is off, which makes it ideal for real-world applications.
The research team explored new advances in the physics of moiré fringes. Moiré stripes are a geometric design that appears when two patterns are layered on top of each other. When 2D materials are stacked, new properties appear that do not exist in a single layer. When these layers are twisted to form moiré fringes, unprecedented tunability of electronic properties becomes possible.
For the new device, the researchers combined two different types of atomically thin materials: bilayer graphene and hexagonal boron nitride. When stacked and purposefully twisted, these materials form moiré fringes. Researchers can achieve different electronic properties in each graphene layer. With the right selection of twists, the researchers utilized moiré physics to achieve neuromorphic function at room temperature.
To test the transistor, the team trained it to recognize similar but not identical patterns. They have also introduced a new type of nanoelectronics device that analyzes and sorts data in an energy-efficient way.
First, the researchers showed the device a pattern: 000 (three zeros in a row), and then, they asked the AI to recognize similar patterns, such as 111 or 101. "If we train it to detect 000 and then give it 111 and 101, it will know that 111 is more similar to 000 than 101," the researchers said. 000 and 111 are not exactly the same, but they are both consecutive three-digit numbers. Recognizing similarity is a higher-order form of cognition known as associative learning."
Technology