THANK YOU FOR SUBSCRIBING
NeuroMem Technologies: AI-Powered Anomaly Detection
When we listen to a familiar piece of music in the morning, we recognize it effortlessly. This is due to an inherent perception of the patterns in the piece of music, which our mind has learned to trace over time. Our ears are so attuned to it that they detect any errors immediately—these are anomalies. This is precisely how neurons work. The same principle of pattern recognition and anomaly detection forms the basis of NeuroMem Technologies. The company provides solutions that allow everyday objects to perceive their environment and interact with users in industrial sectors like security, automotive, aerospace, health, and even at home. “We look at pattern recognition as an ethic that requires little computing or programming—our AI-powered products learn on the go,” says Professor Dr. Pierre Brunswick, CEO of NeuroMem Technologies.
The idea of anomaly detection using Perceptron classifier was introduced in the 1980s but had to be abandoned due to lack of computing technology, big data and the cloud. With the current technological advancements, it rose like a phoenix from the ashes and is now named Deep Learning. Neuromorphic technology instead has been working since late 80’s by simply recognizing symbols and patterns. The company’s products, NeuroMem IP and NeuroMem Chip learn patterns over time and always detect anomalies flawlessly and report them to the user. Powered by AI technology, NeuroMem’s learning and inference engine on a single state-of-the-art neuromorphic chip works as a technology enabler and assists its clientele in designing its own products and POCs, and accelerates the time to market.
NeuroMem excels at iris and face recognition, which is commonly used as a security feature. The neurons detect faces without having to identify them as an individual, which cuts down on the need for personal information in its functioning. Using parallel architecture allows the massive scalability of the product makes it usable in crowded areas like metros and airports.
The automobile industry has also benefited extensively from NeuroMem Chip as its principle of anomaly detection makes it simple to avoid collisions. Dr. Brunswick explains, “We merely ensure that the vehicle travels without any obstruction in its triangle of vision, failing which, the brakes are applied.” The use of anomaly detection instead of computing, coupled with its low energy consumption of 50 mW makes it usable even in case of accidents, low power, and internet shutdowns.
The NeuroMem Chip also tracks the infamous head tilting and the eye movement of sleepy drivers. The stereoscopic vision follows the movement of the vehicle and re-centers the image when required, which provides automatic tracking. This feature also proves beneficial to army drones, as they can be tracked without computing, hence there are fewer chances of the tracking data reaching an unexpected and potentially dangerous source. Drones empowered with NeuroMem’s technology can set their targets on the go and require minimal communication with the authorities on the ground.
In the healthcare sector, wearable IoT devices that are widely used to monitor physical health adapt to the specific requirements of the user over time. They recognize patient history, identify body patterns and issue alerts in case of any changes. The realization that data is an asset for most companies led NeuroMem to avoid store vast quantities of data on the cloud, where it can be easily hacked, but build pattern learning capabilities into the NeuroMem chip. The data is stored locally with encryption devices, so as to secure its access only to authorized personnel.
NeuroMem, in collaborations with organizations in Singapore and Malaysia, works with universities in India, China, and Hong Kong to improve its product offerings. “We are developing programs for countries who want to embark on long AI-powered journeys,” concludes Dr. Brunswick.