Cognitive IoT - the immediate future of IoT

With billions of interconnected devices throughout the globe and their numbers taking an exponential growth everyday - the real concern is, deriving the necessary and required value from the enormous of data generated by them within a small span of time.

Predictions say that by 2020, the IoT network will comprise of nearly 30 billion devices. But today, almost 90 percent of the data generated by these devices go unused. At present, we are merely scratching the surface of the IoT. We are yet to utilize the real potential that IoT has to offer.

Research have pointed out the increasingly symbiotic relationship between the Internet of Things (IoT) and cognitive computing - the artificial intelligence capabilities of the later makes a perfect combination for the speed and size of the former. IoT provides the data quantities needed to optimize the value and Return on Investment (ROI) of cognitive analytics solutions. Such studies are indicative of a trend which will establish the presence of IoT into the data world more strongly.

Cognitive computing can be perceived as a simulation of human thought processes in a computerized model. Cognitive computing involves self-learning systems that use data mining, pattern recognition, vision and natural language processing to mimic the way the human brain works. Cognitive computing systems use machine learning algorithms and deep learning neural networks. Such systems continually learn and acquire knowledge from the data fed into them continuously. In this process, the system learns to accurately refine the way they look for patterns, as well as enhance their methods of processing data. As a result, they become capable of understanding and predicting new problems and modelling their possible solutions.

Cognitive computing addresses complex problems - problems which are ambiguous and uncertain. In today's dynamic and information - rich world, data is of various types and forms. At such a fast paced world, the way users use and interact with their surrounding device has changed drastically. Users expect to interact with the machine in a way they interact with other human beings. Users expect immediate insights from the massive amount of data generated in a fraction of a second. Cognitive computing aids in keeping up with the pace by providing a synthesis of information, influence, context and insights. This requires cognitive systems to evaluate all the available data, algorithms, outcome required and suggest the best possible (not just the right) way to approach an analytical solution to produce the desired insights.

Cognitive computing systems enables computation of the necessary features. It identifies and extract context features such as location, time, task, history or other attributes to present an insight that is appropriate for an individual or a dependent application engaged in a specific process at a specific time and place. Cognitive computing helps in machine - aided methods by walking through massive amount of heterogeneous data, finding patterns to address the requirement of the moment.

Digital devices had become very prominent and has essentially become a part of our lifestyle. Cognitive computing makes the interaction between human and device seamless. They take in all the analog inputs like natural language, visual images and interpret them providing results in context with the help of Artificial Intelligence, Machine Learning, Deep Learning techniques.

WHY IoT needs cognitive computing?

Cognition means thinking, while computers are not yet capable of general human - like thinking, they are slowly being able to perform some of the underlying function that human perceive as thinking. Human cognition involves 3 things - understanding, reasoning, learning.

In computing, understanding means being able to take in large volumes of both structured and unstructured data and deriving insights from it. Reasoning denotes using a particular model to derive answers or solve related problems without having the answers and solutions specifically programmed. Learning denotes being able to automatically infer new knowledge from data which form the basis of understanding at a scale.

Cognitive computing is the perfect match for IoT specifically for certain critical reasons -

The rate and scale of data generation - The massive amount of data getting generated from IoT will soon overwhelm the human ability to analyze them for detecting patterns and learning. Learning will help optimize these connected systems to become more efficient by combining sensor data from the system along with contextual information. Applying advanced analytics methods like machine learning or deep learning becomes crucial to scale the IoT data.

Heterogeneous data sources and types - One of the most common challenge that IoT data pose is the heterogeneity. For IoT, there may be multiple data sources that provides relevant information or context for better understanding and decision making. IoT data needs special attention - it needs the ability to take in and analyze different data types of data, including digital sensor data, audio, video, unstructured textual data, location data etc - to identify patterns and their correlation. Understanding the user inputs and behaviours with their surrounding devices could be enhanced with the knowledge of the context - physical, temporal or even emotional context. Reasoning and decision making could also be improved by integrating heterogeneous data sources, feeding in more data from time to time.

 

Devices’ shift to the physical world - With the growth of IoT, more and more people of different age groups are interacting with connected devices, there is a need to enhance the human - machine interaction, make it more seamless so that human and machine can co-exist. There is a need to focus on more human centric interfaces where people are able to interact with their ambient devices in their native languages much like they would interact with other humans. Cognitive systems would be built in such a way that they can understand human beings.

In a nutshell, cognitive computing capabilities to IoT will enable the timely aggregation of sensor data from multiple sources. The application of AI techniques (machine learning, deep learning etc.) will give rise to intelligent data to feed into the predictive analytics pipeline for fully synthesized, time - sensitive, cognitively analyzed data. The marriage of IoT and cognitive analytics would enhance machine automated actions and human centric decision making. Cognitive computing is certainly the immediate future of IoT - improving the efficiency of complex, sensor - driven systems through learning and infusing more human awareness into the devices and environments we interact with. This will enable connected things to understand and interact with us in our own language(s) without having to press any buttons. It would create and essence of human - aware devices.

Sukanya Mandal
Data Science Professional & IEEE Member