DIHGIGAL, Support and/or implementation of feature selection techniques

To achieve quality data, it is essential to eliminate irrelevant variables, which can be done through feature selection. This service offers support and developement capabilities to introduce advanced techniques for feature selection in order to optimize the amount of data processed for a particular data-related task, generally involving an intelligent or decision-support system.

The emergence of Big Data has revolutionised virtually all areas of our lives. our lives. This phenomenon has been brought about by the exceptional increase in the amount of digital data being produced and stored every day, as well as the exceptional increase in the amount of data being produced and stored amount of digital data being produced and stored every day, as well as the proliferation and commercialisation of ubiquitous devices that allow us to be proliferation and commercialisation of ubiquitous devices that allow us to be always connected and constantly always connected and constantly accessing all kinds of information.

Without However, we sometimes forget that it is better to go for data quality rather than data quantity. rather than quantity of data. And, in order to achieve quality data, it is essential to eliminate those variables that are irrelevant, which can be done through feature selection.

Typically, feature selection is performed on general purpose computers, which work efficiently on a class computers, which operate efficiently in double-precision (64-bit) floating-point and provide RAM memory of hundreds of GB with high bandwidth.

However, other architectures that focus on minimising power consumption by providing fewer hardware resources are now increasingly being used. Examples of platforms where feature selection is offered include GPUs (Graphic Processing Units), with hundreds or thousands of low-frequency cores, or wearable devices (glasses, watches, bracelets or even rings), designed to be worn as a garment or accessory. Efficient exploitation of these technologies requires changes in the implementation of feature selection algorithms.