What has the data revolution really afforded us?

Almost every technological development in the history of mankind has received mixed reactions. Things have not been any different in case of the tremendous improvement in the field of data science and analytics throughout this decade. This article is neither to blindly eulogize nor to criticize the progress. I would rather like to present a story that speaks for itself.

The Cognitive story

I had come across a keynote speech by Tanmay Bakshi a few months ago. He talks about quite a few things regarding the concept and application of deep learning but what really had me was the ‘Cognitive Story’. It is an initiative by IBM, Darwin Ecosystem and Not Rocket Science to use AI and deep learning for the betterment of people with special needs.

The team had gotten a call from a hamlet near Toronto. It was from the mother of a twenty nine years old woman Boo, a patient of Rett Syndrome. She could not communicate either physically or verbally; she could barely make some subtle movements making the broadest concepts understood. Her mother wanted to know whether it was possible to use AI to help Boo.

The team made a fairly long journey and reached their house. And the first chapter of the cognitive story started.

The plan

Whenever Boo tries to communicate with her family through some slight movements her brain transmits EEG signals. The plan was to track those signals to understand her and relieve her from the isolation. Boo’s mother is capable of understanding many of the signs made by Boo. She can tell when she is hungry or when she wants to watch television. Of course, she can do it because of her intimacy with her daughter. The team wanted to use her insights regarding the movements to label the respective EEG signals so as to train the algorithms to analyze those signals.

The challenge

The EEG signals give out 250 data points in a second. Every data point has 16 values. That would mean 4000 values in a second. The long short term neural networks or LSTM neural networks as they are called are meant to deal with this sort of data. But labelling those signals and training the algorithms is clearly a gigantic task. Moreover designing a head gear that can translate the EEG signals into comprehensible language in real time seems to be equally difficult.

How they plan to cope with it

The best in the business are on the project and what is more awesome is that the project has been made open source. Algorithmists, deep learning experts and engineers from across the globe are welcome to contribute to it.

With the success of the project Boo can find a voice. More importantly a development of this sort can benefit thousands others who are dealing with such challenges.

The point of it

Science often moves too fast for us to totally comprehend its intricacies and effects. And like it or not that is how things are going to be. What we can do is keep ourselves updated; move forward; look for chances to improve. If you are sitting in Bangalore, getting tired with your repetitive job, why not start with big data analytics training in Bangalore. You never know what the future may hold.  

Leave a Reply

Your email address will not be published.