Feature Articles

Can computers ever think like humans? Cognitive computing is coming to bridge that

By Vijay Anand - 7 Apr 2016

The future is cognitive computing

 

Day 2 at NVIDIA's GTC 2016, and we see one of the first Deep Learning Robotics application demo with the IBM Watson robot. Freaky, but definitely interesting. #gtc16 #nvidia

Posted by HWM Malaysia on Wednesday, April 6, 2016


The future is cognitive computing

Finally, a deep learning robotics application that many can relate! Of all the deep learning examples we’ve come across, this IBM Watson Robot demo is one of the most appreciated and understandable for the masses.

This is but a glimpse of cognitive computing. But what does that term mean? Is it achieving artificial intelligence? They are similar, but not quite the same. Cognitive systems amplify human cognition.

In today’s keynote at NVIDIA GTC 2016, Rob High, IBM Fellow and CTO of IBM Watson, walked us through some insightful aspects how the data overload is urgently creating the need for such cognitive systems to better make sense of things.

In his words, it’s not about re-creating the human mind. It’s about processing and understanding data to make meaningful decisions. As you would be aware, a single data point doesn’t tell you much, but processing a lot of data helps you pick up a data pattern, trend or even a solution.

Cognitive systems are able to learn their behavior through education and support forms of expression that’s natural with human interaction. Through the process of gaining experience and expertise, cognitive systems can evolve and respond to new information and scenarios to tackle and respond back in new ways.

If you recall, IBM’s Watson enrolled in the popular game show Jeopardy! in 2011 and won. It wasn’t a coincidence nor did the outcome come about by searching for answers. It was possible only because Watson was equipped to process general domain knowledge to learn and compartmentalize through what IBM introduced a process called a factoid pipeline. Since then, here's how IBM's Watson has evolved over time:-

And here are some example of cognitive processing at work:-

Here’s a basic example of cognitive processing at work. While the traditional equivalent of it is through pre-categorization of products, a cognitive processing system backend can take in more open ended inputs to derive the outcome for the customer.

A young child’s toy form Elemental Path can converse and bond with him/her by adapting to the child’s developmental needs. At the same time, it’s able to identify developmental concerns or lapses to parents based on its time of interaction.

But cognitive systems don’t just fall from the sky and it will require a lot more work through new machine learning techniques with deep learning. That’s where NVIDIA’s vested interest and highly advanced GPUs such as the just announced Tesla P100 come in to play to unlock the future potential of cognitive computing and take it to a whole new level.

Deep learning is essential to make the next set of leaps in computing and for machines to understand humans.

An example of a convolution neural network that processes the image to derive what it is.

Taking an example from the medical industry, Rob talked about the plight that doctors face today. Each human is unique and so are some of their illness. There is a need to create a more personalized treatment for each individual. Cancer for example could have countless variations and might be as unique to each individual. There may not be one generic solution to cancer. Such focus require doctors to read a lot of medical literature to derive something useful. In reality, they don’t really have much time. If there was a cognitive system to help doctors crunch through all the available literature, this will enable doctors to then process the research output to better administer a suitable medication for their patients. They can then deliver better treatments and improve their patients’ wellbeing.

Rob thinks that in a decade or earlier, there will be a lot of cognitive systems learning and interacting with humans, just like how transaction processing is prevalent today. To accommodate this, there will eventually be a fundamental change of systems interaction in the near future, similar to how the mobile phone facilitates us in our everyday tasks.

Exciting times ahead and we can’t wait to see what new paradigm shifts will take place in the coming years.

Join HWZ's Telegram channel here and catch all the latest tech news!
Our articles may contain affiliate links. If you buy through these links, we may earn a small commission.