Data is fueling the insight economy and has the potential to disrupt entire industries. According to the IBM Research Global Technology Outlook, a new approach is required in order for businesses to benefit from higher quality insights and time-critical decisions.
The next IT paradigm
The rate of data creation is twice the expansion of bandwidth over the past decade. It takes 3 hours to transfer 1TB on a 1Gb link. Transferring 1PB of data over the same link would take 115 days. Given the fact that 60% of valuable sensory data loses its value in milliseconds it is obvious that attempting to move data is not the way forward. The solution is to move the compute part to the data.
Brain-inspired cognitive system
The analytics continuum will move from descriptive (discover, report, analyze) via predictive (predict, decide, act) to cognitive (understand, reason and learn). One of the new IT platforms that will emerge on the edge is the brain-inspired cognitive system (a neural network). These systems learn their behavior through education and continue to evolve as they experience new information, new scenarios, and new responses. They excel at recogizing complex and subtle patterns and are at the moment, primary used to analyze sensory data like recognizing objects in a video for example.
IBM’s SyNAPSE implements these neural networks in neuromorphic hardware. With only 256 neurons available in 2011 and 16 million neurons available today, it is a matter of time before the new possibilities for analyzing data will be available to more than just sensory data.
Have you moved your “compute” to your data yet? What kind of data would you analyze with a cognitive system? Let me know in the comment section below.