5 areas for making cognitive computing robust

5 areas for making cognitive computing robust

Voice activation, visualisation, and context awareness need to be further implemented so interaction feels more natural to humans, IBM says

IBM has published a study that outlines five focus areas where advancements in cognitive computing could be made to make it more robust in future.

Cognitive computers are self-learning systems that simulate how the brain works and use machine learning algorithms to do data mining, pattern recognition and natural language processing for solving complex problems. IBM is a leader in the cognitive computing space, with its Watson supercomputer and SyNapse, ‘brain like’ chip.

"While tremendous advancements have been made over the past 50 years, cognitive computing is virtually in its infancy in terms of how this exciting technology could potentially evolve," IBM said in its study.

The first focus area looks at personalised and interactive features of cognitive computers. The problem with many cognitive computers today is they are passive in nature in that they still require humans to action an output, IBM said. For example, humans still often interact with cognitive systems through typing on a computer, mobile or Web portal.

Voice activation, visualisation, and context awareness need to be further implemented so that interaction feels more natural to humans, the study said.

The second focus area is the degree of autonomy in learning. "Current cognitive systems are predominantly trained systems (supervised learning). These systems rely upon humans with domain-specific subject matter expertise to train them. This process can be more labour intensive and time consuming," the study said.

Greater implementation of unsupervised learning – which is more exploratory data analysis rather than training a system on given labels to make predictions – will allow for less human interaction in the process of training the systems. "The research community is actively looking to make breakthroughs in this area," the study said.

The third focus area, the various types on inputs cognitive computers can sense and interpret, focuses on natural language processing (NLP). Today, NLP is still predominantly text based, said IBM.

According to the study, cognitive systems in future will accommodate a variety of media, not just text – for example, audio, image and video.

The ubiquity of cognitive computers looks at how widely accessible this technology is, whether it be over Web portals, mobile apps and cloud. With increasing adoption of this technology, cognitive computers will become ubiquitous in future, IBM said.

"This future could include a marketplace with millions of cognitive agents or avatars, driven in part by the explosive adoption of mobile devices, the IoT [Internet of Things] and the upsurge of machine-to-machine interaction.

"Tomorrow’s cognitive computing fabric will be interwoven into technology (such as social media), thereby touching our daily lives," IBM said.

Scaling capability to meet demand is an important focus area to making cognitive computers robust.

In 2011, Watson required 90 IBM Power 750 servers. By January 2014, Watson was 24 times faster, had a 2,400 per cent improvement in performance and was 90 per cent smaller, said IBM.

Follow CIO Australia on Twitter and Like us on Facebook… Twitter: @CIO_Australia, Facebook: CIO Australia, or take part in the CIO conversation on LinkedIn: CIO Australia

Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.

Join the newsletter!


Sign up to gain exclusive access to email subscriptions, event invitations, competitions, giveaways, and much more.

Membership is free, and your security and privacy remain protected. View our privacy policy before signing up.

Error: Please check your email address.

Tags IBMcognitive computingcognitive systems

More about Facebook

Show Comments