NetObjex Blog.

Where Cognitive Science Meets IoT

Created on February 19, 2016

Cognitive Science, as a discipline, is fascinating in the sense that it tries to understand thinking and intelligence amongst other things. Engineers for the longest time have tried to quantify and codify human intelligence into a series of processes in the hopes of making intelligent devices.

Enter the world of the Internet of Things (IoT) or smart, connected devices. Are these devices really smart? Well, lets go back to Cognitive science and examine some of the concepts informally. What are some of the building blocks of thinking.

  • Memory: One needs memory for sure because without memory, you can’t store what you have learned in the past and be able to build upon that.
  • Recall: No use having memory without the ability to recall those memories. Like a search engine, our brains have an ultra powerful search engine with instant recall capabilities.
  • Compute: We can compute things. For instance, we know if a loaf of bread costs $1, two loafs will cost $2 by simple arithmetic.
  • Stimuli: We react to external stimuli e.g. we have five senses – touch, smell, sight, hearing, and taste. Given these senses, we have receptors that communicate what we sense and deliver that message to the brain.
  • Learning: Our brains are able to learn new things such as procedures and/or skills. For instance, if we see a new species of fish at the aquarium, we learn to identify this if we see it again elsewhere. Similarly, if we are taught to cook a new recipe for preparing a dish to eat, we are able to learn the steps involved.

Now, let’s try and see how these five building blocks fit within the IoT paradigm. Devices and computers have had memory chips for a long time – both Read Only Memory (ROM) and RAM (Random Access Memory). This memory is implemented in a number of different ways – SD cards, flash drives, SSD drives, disks, and more.

We have the capability of accessing the data contained in storage either through direct retrieval methods or by developing complex search algorithms to query mechanisms in databases to retrieve what we seek.

If you have a Central Processing Unit in a device, then that adds the capability of computing. Many IoT devices come equipped with some limited computing capabilities. The CPU of an IoT device is often called its Controller.

Similar to a human, IoT systems usually come equipped with several sensors which capture data from the surrounding devices. Sensors capture the data and pass the data to Controllers.

Next comes the most important topic – Thinking. Can IoT devices actually think? Are they really smart? Most of the currently available devices simulate “smarts” and “thinking”. A thought process can be codified in the form of a decision tree (at least for those of us who don’t have convoluted logic and not suffering from bouts of emotional melodrama). This decision tree was created by man to deal with the behaviors of a device given all the external stimuli.

So, many IoT devices are not really smart per se. Now, in order to make them truly smart, they would have to develop a capability called machine learning. The device would have to learn lessons experientially and either form new decision trees on its own or provide inputs to existing decision trees.

For example, a sprinkler system being programmed so that it waters the grass every day at 6 a.m. A smart sprinkler system may have a sensor to capture moisture in the soil, thus expanding the human decision tree by adding a new clause – water the grass at 6am provided soil is not wet. This rule will conserve more water but what would be the point of watering the grass at 6am after a rainfall at 5am! So this form of adaptive learning and expanding upon a human decision tree could be one of the ways in which IoT and Cognitive Science can be combined to arrive at achievable and effective solutions.

The NetObjex IoT platform is a state of the art platform for devices of all kinds and provides a framework for machine learning while handling all aspects of device management including Fleet management, Rules management, Enterprise Integration, Campaign Management, and Real Time Location Tracking.

Related Articles