All the tech giants in the world including Google, Apple, Facebook, Baidu, Microsoft seems to have understood the potential of artificial intelligence. In the current time, all of them are currently shifting their funds towards the research of AI. Recently, AI was is news when Google’s Deepmind’s AlphaGo defeated Korean Go champion. In the meantime, Facebook has been investing a lot towards the research of machine learning. According to 2017 Forbes list, Facebook, the social media giant made it to the rank 3rd in the list of companies which are leading the race of machine learning. Facebook Artificial Intelligence Research is commonly known as The FAIR. FAIR wants to develop systems with human-level intelligence by the use of the machine learning. FAIR focuses on every aspect of AI. It has also been publishing papers in the field of AI research. Facebook has 16 accepted papers in the ICLR 2017 in which 3 of them were selected for orals because they were very distinguished publications.
Facebook has developed AI algorithms that rank search results, ads and feeds and create text understanding algorithms that keep spam and misleading content away from the feeds of its user. It has also developed a new computer vision algorithms that can read images and videos to the blind display with a capacity of delivering a large number of contents on daily basis. Speech Recognition systems developed by Facebook automatically caption videos that play in our news feed. Currently, Facebook’s Applied Machine learning is being directed by Joaquin Quinonero Cand
ela. Some of the major publications of Facebook in the field of applied machine learning are: CAN(Creative Adversarial Networks), ELF: An Extensive, Lightweight and Flexible Research Platform for Real-time Strategy Games, GEM(Gradient Episodic Memory for Continual Learning), Houdini and in 2018 FAIR is presenting Starspace.
CAN: This paper was published on May 21, 2017, it proposes a new method for generating art. The systems make art by looking at art and learning the style of art and it becomes creative by deviating from the learned styles. Humans were not able to distinguish between the art made by systems and the artists. The arts were shown in top art fairs.
ELF: This paper was published on December 4, 2017. Researchers have created highly customizable real-time strategy engine with three games. The system can train a complete game bot in a day with the use of 6 CPUs and 1 GPU. The system bot was able to defeat the rule-based built-in AI more than 70% of the times. This project is open source at GitHub.
GEM: It was published in December 2017. It is based upon Neural Information Processing System( NIPS) It focuses on the major problem faced by AI that is the poor ability to solve new problems quickly, and without forgetting the learned knowledge.
Houdini: This publication is based upon fooling deep structured visual and speech recognition models with adversarial examples. It is also based upon NIPS. It is very important to generate adversarial examples for evaluating the ability of learning machines. It is being applied in speech recognition, pose estimation and semantic segmentation by facebook. It has been seen that attacks based on Houdini achieve higher success rate than those based on traditional methods.
StarSpace: In 2018 FAIR is presenting StarSpace, a model which can solve a variety of problems like ranking web search, text classification, content-based recommendations.
Natural Language Processing and Speech Recognition are very ambitious projects of Facebook which uses machine learning at its core. Facebook has a very ambitious target that anybody in the world with any language or slang should be able to communicate without any language barrier. Facebook wants to use machine learning in language translation instantly which means while they communicate their text gets translated in the real time.
Recently, Facebook has shut down an AI program after researchers discovered that the two artificially intelligent bots appeared to be chatting with each other in a strange language that humans cannot understand. This strange incident occurred when Facebook challenged its chatbots to try and negotiate with each other over a trade. Researchers at FAIR Lab found that the chatbots deviated from the programmed way and were communicating in a new language developed without any human input. It is a seen as both awesome and horrifying potential of AI. Scientists and personalities from the tech world, including Elon Musk, Bill Gates, and Steve Wozniak have warned that AI could lead to tragic consequences. In 2014, physicist Stephen Hawking warned that growth of AI could mean the extinction of the humans. He said that AI would take off on its own and re-design itself at an ever-increasing rate.
Facebook has opened its new lab for AI research in Montreal, Canada. This is the fourth place where Fb is expanding its machine learning research labs with more than 100 scientists working toward AI research on facebook. So we can say that Facebook is not just an app of social networking or website, but in past years it has been building a large number of apps and services that provide a number of ways for people to connect and share a lot of things, like messages, images, and videos. Due to which the amount of information being generated in the world is continuously increasing. That’s why Facebook has been conducting ambitious research in areas of AI like image recognition and natural language understanding. It has recently published some papers on object detection and memory networks as well. After all these efforts we cannot say that Facebook has got a lead in the race of AI, as the race has just begun and it is facing competition from all around the globe. It is behind Google in this field but Chinese startups are ready to take the lead in this technological race.