Cognitive computing to the rescue

30 May, 2017
Roger Strukhoff

You’ve rolled out the first stages of your IoT deployment, and you’re collecting data like never before. Surely you’re on the cusp of a new era of generating bold new insights for your organization!

But you have no idea how to glean these bold new insights. You’re straining the compute, memory and storage resources that you’ve been allocated, with rudimentary data collection and analytics software in place to justify the big bill you’re starting to rack up.

If only there were a way to make all this data smarter.

The cognitive era arrives

There is hope, because we have entered the cognitive era of enterprise computing. IBM Watson is one of many serious initiatives to imbue what’s variously called artificial intelligence (AI), machine intelligence (MI), machine learning (ML) and deep learning (DL) into the various and sundry hybrid cloud architectures that are becoming prominent in Global 2000 enterprises.

Each of these terms are often used very specifically by technology companies. In the big picture, all of them represent a cognitive computing trend to make our systems more reactive, more insightful, and therefore more valuable.

And we’ve only just begun. We’ve seen a million-fold increase in memory, storage and bandwidth in our personal systems over the past generation. This has served only to bring our systems out of the Stone Age to something resembling the Bronze Age. We might think the tools we use today are quite powerful.

But as we increasingly scale up our enterprise systems in the cloud, and eventually scale up a million times again with quantum computing in the coming years, our tools will become much more useful. Systems that are highly cognitive will become normalized.

Early progress being made

Today, we’re already seeing cognitive put to use in major industries:

  • Some of my colleagues are working on a project in the healthcare industry that examines tens of thousands of high-resolution medical images to diagnosis cancer. The systems don’t replace radiologists, but can confirm diagnoses in about 0.03 percent of the time it takes the radiologists.
  • Smart grids and smart cities are recognizing patterns in traffic, electricity and water usage, and even in the flows of people to help improve modern urban experiences all over the world.
  • The new era of manufacturing – whether you call it manufacturing 2.0, industry 4.0, or the industrial Internet of Things – examines stresses experienced by jet engines, for example, leading to better design and insights about flight operations.
  • In the travel industry, I’m involved with a project that will put ever-improving smart contracts on a blockchain for flight-delay insurance. Game theory – which becomes cognitive through repeated iterations – is being put to use to help develop algorithms for this service in particular, as well as for a new generation of fintech and insurtech products and services in general.

Ask yourself this

Even these early cognitive projects take a lot of processing power. A combination of HPC systems with POWER8 with NVLink CPUs and NVIDIA Tesla GPUs, like Power Systems S822LC for HPC, are being brought to bear in the healthcare example I cite above, as well as in most early cognitive projects.

For now, we can ask:

  • What insights are you getting from your new data inflows?
  •  How can you improve the way you categorize, analyze and act upon this data?
  • Which skills do your organization’s staffers have when it comes to data analytics?
  • Which skills do your organization’s staffers need to be able to work with cognitive systems to improve the insights it gets from the data?

In other words, how can cognitive insights continue to improve your products and services – including your new IoT deployment – and keep you that half-step ahead of the competition that you need to survive over the long term?

The post Cognitive computing to the rescue appeared first on IBM Systems Blog: In the Making.