Machine learning

About the author : Diana

A Valuable Tool in the Future of Software

Software development has been vital in the evolution of the computer industry, with tech companies competing against each other to figure out whose software, services, developer frameworks and user experiences are the best. There are two emerging categories they are attempting to establish dominance in: augmented/virtual reality, and machine learning.

There has been a great amount of focus on the prospects that augmented reality and virtual reality have for both gaming and productivity, including apps like Pokemon Go, Ikea Place, and Glasses by Warby Parker. All of these place virtual objects within the real world, having to track object positioning, motion tracking, scene rendering and, in some instances, lighting conditions. All these exciting visuals seem to have drowned out the exciting advancements in machine learning. From an engineering perspective, both are very impressive. However, machine learning is one of the most useful areas of development in computing today.

Early Machine Learning

The concepts of machine learning have been present for at least 20 years in devices that utilized handwriting recognition. However, during that time, the technology had not evolved far enough to be considered decent.

About a decade later, virtual assistants showed promise for machine learning, creating a means of speaking to a device to get it to do what a user wanted without having to type things out. Though still not perfect, massive improvements were made in the decade that followed.

Machine Learning Today

Machine learning has become an important part of many digital services provided by such companies as Apple, Google and IBM. It is very useful for image analysis to allow a user to search for items in an image as well as to analyze the environment in real time while taking a photograph. Additionally, it is good for digital assistants to suggest applications to open, text to write, reminders of events based on time and location, and analysis of emails and messages for things to be automatically added to a user’s calendar. It is also this technology that allows robots to play games and be trained to handle different tasks, training themselves to adapt to changes in their originally programmed situation. This is used by Face ID, tracking changes in a person’s face to allow the phone to still be unlocked, despite the face looking different.

Machine Learning in Apps

As part of the goal of expanding machine learning, both Apple and Google have begun to allow third party access to their machine learning frameworks. These allow users to design apps that can analyze images, or utilize new sensors to better analyze faces for mapping a mask or glasses to it. Machine learning requires some knowledge of data models in order to cue the computer into what to be analyzing, and how to iteratively update the model in order to improve performance. This ability to adapt is important, since it aids in anything relating to data analysis, from the core of our augmented reality apps to the rise of connectivity between all our devices.