Facebook Highlights Its Contributions to AI In 2018

As shared by Nick Lovrien, the Chief Global Security Officer of Facebook, in the Facebook world, “a day is a week, a week is a month, a month is a year.”

What he means by these words is that development at Facebook moves quickly. Facebook has put Artificial Intelligence (AI) to use for a plethora of applications.

From protecting Mark Zuckerberg to connecting to your brain waves and identifying porn or keep its employees safe, Facebook is one of the many conglomerates which has gone all out in embracing the possibilities of AI and coming up with innovations in the field.

Facebook’s Belief in Artificial Intelligence

2018 was a grand year for Facebook where it contributed to the world of Artificial Intelligence significantly and in a multitude of ways. Facebook has always stressed the importance of Artificial Intelligence learning just like humans in order to bring people together.

This core belief has led Facebook to inspire its engineers in order to create some cutting-edge tools and develop a system that can learn through real-world data.

Highlights of Its Contributions to AI

The Facebook AI Research (FAIR) group aims to develop systems that possess the intelligence of humans. Even though this milestone needs years of work, 2018 was a good effort in that direction as Facebook explored a number of approaches to restrict the need for supervised learning by systems.

In this aspect, Facebook also attempted to demonstrate projects that highlighted how it could be beneficial for us to have systems learning from semi-supervised and unsupervised data.

Another manner in which this social networking giant contributed to AI was through its work on enhancing the number of languages that could be understood or translated by their systems.

It introduced a new technique where their systems could be trained in automatic translation using MT models without any supervision.

[see-also]

Thanks to this research, Facebook was able to add 24 additional languages to its automatic translation services in 2018.

Facebook also introduced something called Omni-supervised learning last year, which was a distinct variation of data supervision.

In addition to that, the company also indulged in a study on hashtag-based image recognition by utilizing Instagram hashtags posted by users for its AI image recognition research.

The results of the study proved that the utilization of countless data points could potentially help in image-based tasks. It also shed light on how hashtags could go a long way to helping computer vision systems recognize the additional elements as well as sub-categories in an image.

In addition, Facebook also updated its popular deep learning framework PyTorch, which now provides additional support to new developers and help them turn their AI experiments into production-ready applications.

The company also implemented AI-based features to identify languages in posts that express suicidal thoughts.

The social media giant also marked its presence in medical research through the development of fastMRI, an effort to improve the diagnostic imaging technology and expedite MRI scans.

Artificial Intelligence has basically permeated all domains, industries, and organizations today. Facebook’s work in the field of Artificial Intelligence has been exemplary, and it has grand plans for 2019.