Rosetta AI filters the impertinent memes screening task in Facebook

Facebook needs no introduction to the public as it is the most used social network all over the world and most successful tech giant in US is now making step further to detect the offensive sprays against public to combat such kiddish actions to keep it clean and polite. As the social network has its more no. of fresh users getting signed up but also there are hasty things among-st users when it comes to comment or share on certain topics brings heat and raging between them causing a deep distress.

Countless no. of texts, videos and images are been uploaded in social media network like Facebook everyday but not all the media that is been uploaded by user are seemed to be found polite or decent . That is not manly possible for human to check out billions of pics uploaded at the same time. Also, companies like Facebook , google relying on artificial intelligence to eradicate the spam and other cause course of problems

cleaner cleaning behind the facebook logo

As Facebook is straitening its measures by taking out fake news more accurately which is wide-spreading like fire with the help of ai and wikiturbine  and also tested their spam detection contents or comments in groups and pages and also in instagram through this deep text learning. Now Facebook has to come over to the biggest struggle by separating out texts from images which we call as MEMES through optical character recognition machine learning system called Rosetta is soon to be stepped up in action for all users

Facebook and other popular social network is making some actions strictly mandatory to the public to make sure that hat-redness are not spread among st users which makes audience to pull off from the social account or getting into war of words unstintingly. Content policy from Facebook has beefed up and screen readers are tested in more accurate manner through modern means with the help of ai.

This Optical Character Recognition technique needs lot of training regularly to process and re-process again to maintain output efficiency to more precise scale. Rosetta not only adopts from images and texts which stored or learnt during training process a lot but also it adapts and extracts tons of billions of images and video frames in unbelievable wide variety of languages every time roughly mastered up to filter out such memes.

In their recent post , they explained how it functions : It gets a start by detecting rectangular regions in images that contains text. Then with the help of neural network training and based on R-CNN these detection are performed.  it checks out whats it is been written in the transcribe in any region say Arabian , Hindi , German and its been tested out both manually and automatically with the help of pre-training and also with the manual automated indulgent. It is achieved with the help of Facebook extraction of human and machine interactive annotation images. Now it seems that only detects the words in English but also many languages very carefully through bounding box regression [ core by core and frame to frame detection ]

image extraction technique used in facebook to test

    ( Image Source : Facebook code team )

Group of team from Facebook and for Instagram are testing this feature to make sure the precision is to close to the state of art technique to keep the content clean and policing the reforms for better efficiency and efficacy and also testing in different languages to keep it grow. On the top of these, It has added 24 languages to the translation services to do these tasks .

Architecture of the text recognition model process
As you can see from the above architecture model that how the image and learning been processed very sequentially with the sequence loss of CTC which is very harder to train them than conventional training model adoption. As we all know that Text Recognition Model is mostly used for testing English or Latino data sets but now the Facebook has taken steps even further with these hardening extraction to make the complexity more simple in future. But right now , Facebook is more dependent on Human moderators to vigilate such roles to keep it off the bay and AI being infant to understand such memes or video in the same way the person would. However, based on progressive regression model it has quiet improved a lot and have to wait for the future accuracy whether it will be doing it more accurately or will it scan all the images and put those under fallacy mode is a rising question too.

Check out Related Posts : US Citizens can send and pay using fb , US Citizens can now order foods via. fb, Facebook testing its Face Recognition feature, WhatsApp embedded in Facebook

Robots in Japan are now training English to boost linguistic skills

Japan is a well developed and market oriented economy and in Asian countries, it has its ace in the field of advanced robotics for a decade till date. Japanese education system is distinctive in all accords with their traditional approach but never ending updation in automation from them has spurred a lot which enhances not only education but also technology and also it's well known for the international conglomerates such as Sony, Panasonic, Fuji , NEC , Nintendo , Epson , Fujitsu , Hitachi, Sharp and Toshiba.

As we know the education system has impounded technological presence in day to day activities with the help of STEM ( Science , Technology , Engineering , Maths ). The usage of technology and hands-on-learning in schools for primary , secondary and higher secondary has never gone belittle. Though STEM is significant in playing academic education English has still pre-dominant worldwide.

So, Japanese education system has taken steps to induce English in primary , secondary and higher secondary schools through self mentoring and also with the help of AI & Robotics. Japanese Ministry of education pushes the necessity of English in schools with the help of English speaking robots roughly around 500 schools all over Japan. It seems they now feel the necessity of learning English to compete global scale in linguistic exchange for mutual benefits and we could expect study apps and online conversation sessions with English mentors and speakers like in BBC learning English forum to educate , build vocabulary and improve verbal and oral communication skills in English among-st students from early childhood 

japanese robot teaching english

( Image Broadcasting source : NHK world )

You can see the range of programming movements and also with the hand gestures it trains English to the students in easy understanding way. NHK told the reporters that Japanese students are not good in either speaking or writing in English and curriculum guidelines that will be implemented in schools to nurture such skills in primary level to improve their traits through robotic interactivity. In 2009, Japan tried a robotic teacher called Saya that is an humanoid robot which takes as a role as school teacher in elementary schools which was successful.

Schools in Japan are in now compulsory to implement such robots to take in-charge of English training more vigorously leaving real time human teachers for other subjects and no wonder Japan is always a way ahead in taking imperative measures in larger scale for every institutional use. No. of schools around Japan have explored whether robots can be helpful in more deeper level and also smartening children through most interactive classrooms will make the students more visually appealing and understanding lessons.