During the past 5 years, machine learning technology has found way into applications that see daily use. Examples include Google’s search and image recognition, Facebook’s facial recognition, Apple’s “Siri” digital assistant, and other speech recognition applications.
Especially Machine Learning discoveries with digital images (w. pixel data: Red, Green and Blue values 0–255) has gained some attention for pushingcomputer vision abilities to new heights and for spawning images with seemingly artistic and inexplicable features.
Machine learning softwares can categorize, classify and recognize objects, phenomenons and people’s faces in an image, sometimes to an uncanny extent. A good example is Facebook’s relatively new feature which discovers which friends to tag in the picture you are about to post — if they can do that, what else can they do? Can you distinguish the mix of dog breeds one dog may have? Microsoft can. How is it possible?
First of all it requires a large data source — Tens of thousands of images — to be able to train a “model”. Training a model means running thousands of certain statistical analyses, to be structured and serialised and saved. Upon feeding the ‘learned’ model an image, it will recognize, based on the type of model and training, what is in the image
The first notable machine learning image classification program was discovered and written by Yann LeCun in late 1980’s. It was able to recognize numbers in B/W images of handwritten digits despite large variability. It was so accurate that a system LeCun helped develop, “reads an estimated 10 percent of all the checks written in the US”.
Unity/Unreal Developers, Full Stack Devs, Computer Engineers, Computer Scientists, User Interface Designers, Project Managers, Strategy Director, Account Managers
Always looking for:
Creative Technologists, oF/Cinder/Unity C++ Creative Coders, & Producers
Carreer opportuniteis, firstname.lastname@example.org
For Intern opps, email@example.com