Contact us

Big Data: How Computers Can Learn

Big Data: How Computers Can Learn
Big Data: How Computers Can Learn

Large enterprises, research centers, cities or public entities, all are surfing the Smart City wave; different players with different conceptions of the same topic. However, the Telecom idea, based on Internet of Things, Big Data, Visualization, Machine Learning and even Deep Learning, is quickly spreading due to the expectations for its Return on Investment.

Ajit Jaokar is an expert on this Telecom vision of a Smart City. He came to the MCS to explain how the Internet of Things may improve citizens´ quality of life.

This concept of Smart City using Big Data in city services may be split in three layers:

  • IoT Data. IoT is all about Smart Objects. An object is Smart when it has an identity, a communication mechanism and a set of sensors/actuators. A sensor collects data in a standardized language and this raw data is sent to the cloud.
  • Cloud. It stores and transforms raw data into useful information.
  • Service. The data gathered from IoT and transformed into useful information interacts with the service in real time to improve it. But, what is Big Data

But, what is Big Data?

Big data is a dataset with 5 attributes based on Vs: high Volume (terabytes), high Velocity (real time), high Variety (from different sources: video, audio, text or image among others), Veracity (real data) and Valuable data.

In the urban scenario, Big Data means all the data gathered via surveillance cameras, smart public transport cards, social networks, credit card payments or utilities, among others, that may be analyzed and used to understand urban complexity and predict future patterns. However, the task of obtaining information from this data is not feasible for humans in real time. Thus, computers must be trained to extract this information, using machine learning techniques.

An example of this philosophy is NEST. NEST is a Google-produced sensor/actuator for smart homes that uses machine learning to understand user´s behavior patterns and accordingly adjusts the household features (temperature, lighting…), thus becoming better with use. This is the philosophy of a Smart City; automatically adjust urban services according to citizen’s needs.

And, what is machine learning?

Machine learning is based on programmable devices that automatically improve with experience. As an example, Facebook has used machine learning with photos. For many years, people have been tagging friends, so nowadays the Facebook software is able to recognize a set of pixels as the shape of a face.

Machine learning depends on a training dataset. After a proper training, the algorithm is able to make predictions on its own. As it is a statistic science, predictions can never be totally certain, but they get better as the number of cases studied increases.

In complex systems as cities are, predictions depend on a large number of processes. Given the complexity of the models, programmers have developed deep learning.

Finally, what is deep learning?

Deep learning is a type of machine learning algorithms that use a cascade of many layers of nonlinear processing units for feature extraction and transformation. Each successive layer uses the output from the previous layer as input.



Leave a Comment

  • Universidad Politécnica
  • ETSI arquitectura
  • ETSI Industriales