TrueNorth was primarily developed in order to simulate neuronal networks, and can be used to recognize images and patterns. The hardware offers numerous application possibilities for sensors, the Internet of Things and supercomputing. The chip’s operation is modeled after a simplified model of the neocortex – the multisensory part of the cerebral cortex. A major advantage of TrueNorth is its low energy use: with 70 milliwatts, it needs up to 10,000 times less electricity than other processors that have been designed for use in the IoT. As a result, TrueNorth can be used in even the smallest devices.
The brain-emulating approach behind TrueNorth follows a similar logic as with the large-scale project Watson, whose current state of research can also be seen at IBM’s stand. This cognitive software system is not programmed using rigid algorithms; rather, it analyzes interactions and experiences in its surroundings and educates itself further completely independently. According to IBM, this is important because 80 percent of the data that is produced today is restructured and can hardly be used by non-adaptive systems.
Watson can read millions of such data per second, analyze the data, and understand it. For example, it would be able to recognize patterns in the financial transactions of banking customers. Two other application examples are making conclusions about the purchasing behavior in retail or determining how telecom services are used. Cloud-based APIs are available to integrate the cognitive technology into a company’s IT.
These and other showcases and demos can be seen at stand A10 in hall 2.