Engineers at Northwestern University have achieved a major breakthrough in making artificial intelligence (AI) more energy-efficient. Their newly developed nanoelectronic device can perform machine-learning tasks using 100 times less energy than existing technologies. This leap in efficiency means that the device can process large data sets in real time without needing to send the data to the cloud, dramatically reducing both energy consumption and latency.
The device's small size and minimal energy requirements make it ideal for inclusion in wearable technology such as smartwatches and fitness trackers. This enables real-time data analysis and quick diagnostics, paving the way for more timely medical interventions. For instance, the device can process and analyze electrocardiogram (ECG) data on the spot, negating the need to send the data to remote servers for interpretation.
In a series of tests, the researchers used the device to classify 10,000 samples of ECG data. The results were highly accurate, identifying six different types of heartbeats with an accuracy rate of nearly 95%. The device's capability for quick and accurate classification holds significant promise for the medical field, particularly for diagnosing and treating cardiac issues.
Traditional machine-learning devices require a large number of transistors and significant energy to process data. In contrast, Northwestern's device uses just two reconfigurable transistors made from a combination of two-dimensional molybdenum disulfide and one-dimensional carbon nanotubes. This innovative design allows the device to dynamically switch between various data processing steps, offering both energy efficiency and a compact form factor.
Beyond the immediate benefits for healthcare and wearable technology, the device also enhances data security by processing information locally, reducing the risk of data breaches. The researchers believe that these energy-efficient nanoelectronic devices could eventually become a standard component in everyday wearables, optimized for each user's health profile. The study, supported by the U.S. Department of Energy, National Science Foundation, and Army Research Office, was published in the journal Nature Electronics.
Glossary
Artificial Intelligence (AI): A field of computer science focused on creating machines capable of performing tasks that normally require human intelligence, such as visual perception, speech recognition, and decision-making.
Nanoelectronic Device: A microscopic electronic component, typically smaller than 100 nanometers, used in computing and data processing.
Machine Learning: A subfield of AI where algorithms enable computers to learn from and make decisions based on data.
Cloud Computing: The delivery of computing services—including servers, storage, databases, and more—over the internet to offer faster innovation and flexible resources.
Latency: The time it takes for data to move from one point to another, often experienced as a delay.
Wearable Technology: Electronic technologies or computers incorporated into clothing or other items worn on the body, like smartwatches or fitness trackers.
Electrocardiogram (ECG): A medical test that measures the electrical activity of the heart and is commonly used to diagnose various heart conditions.
Arrhythmia: An irregular or abnormal heartbeat, which can be harmless or potentially life-threatening.
Transistor: A semiconductor device commonly used to amplify or switch electronic signals and electrical power.
Two-Dimensional Molybdenum Disulfide: A material made up of molybdenum and sulfur that is used in semiconductors, often in extremely thin, two-dimensional layers.
One-Dimensional Carbon Nanotubes: Nanoscale cylindrical tubes made of carbon atoms, known for their strength and electrical properties.
Data Breach: Unauthorized access and retrieval of sensitive information, often leading to the compromise of user data.
U.S. Department of Energy: A federal agency responsible for overseeing the United States' energy policy and research.
National Science Foundation: An independent agency of the U.S. federal government responsible for promoting scientific research.
Army Research Office: An element of the United States Army responsible for funding extramural basic research.
Nature Electronics: A scientific journal where the study was published, focused on the field of electronics.