OK
FOR THE BEST EXPERIENCE USE GOOGLE CHROME
OR UPDATE YOUR BROWSER.
LOADING
TURN UP YOUR VOLUME
X. PX
Y. PX
08:16:09
SUNDAY, 27 MAY
IP ADDRESS: 54.162.163.181
BROWSER:
UNKNOWN
VERSION:
?
SYSTEM:
UNKNOWN
TERMS AND CONDITIONS
/
GR
_PAUSE
SOUND _ON
_SKIP
BACK
CHOOSE A PHOTO FROM FACEBOOK_
ΕΠΙΛΟΓΗ
SOUND _ON
LOGIN
HELLO
THIS IS THE ROBOT
WITH THE MOST ADVANCED
ARTIFICIAL INTELLIGENCE_
SHE KNOWS EVERYTHING.
EXCEPT WHAT LOVE IS_
SHOW HER WHAT LOVE IS
BY UPLOADING A PHOTO
OR A TEXT_
HELP HER LEARN WHAT LOVE IS_ _SEE WHAT SHE HAS LEARNED SO FAR
SEE HOW ARTIFICIAL INTELLIGENCE
ANALYSES YOUR INPUT.
LOGIN
WITH YOUR INFO_
FACEBOOK LOGIN_
GOOGLE LOGIN_
SHOW WHAT LOVE MEANS TO YOU_
_UPLOAD A PHOTO
FROM YOUR COMPUTER
UPLOAD A PHOTO
FROM FACEBOOK
or
IF YOU DON'T HAVE A PHOTO, WRITE A FEW WORDS HERE:
THANK YOU.
YOUR IMAGE, ONCE APPROVED,
WILL BE UPLOADED TO THE DATA CLOUD_
_ELLi ANALYSES
YOUR PHOTO
_ELLI ANALYZED YOUR PHOTO
AND KNOWS THAT IT SHOWS:
LABELS
_KISS 100%
_LOVE 99%
_HUG 95%
_KISS 100%
_MOVIE 89%
FACES
_JOY
_SADNESS
_ANGER
_SURPRISE
COLOURS
SEE HOW ARTIFICIAL INTELLIGENCE
ANALYSES YOUR INPUT.
BACK
_HOW WE ANALYSE YOUR INPUT
WITH THE HELP OF ARTIFICIAL INTELLIGENCE

ELLi is a project that gives a more human-like face to Artificial Intelligence. Our goal is to help her give an answer to one of the most popular questions people ask Google with the use of Artificial Intelligence.

To achieve this, we take the photos and the texts that represent what love is to you, and analyse them through automated learning tools, in order to identify patterns and common characteristics.

In this stage, Google’s Could Vision API uses the most advanced Artificial Intelligence to identify what is shown in each photo. Next, the elements of each photo are automatically translated by Google’s Cloud Translation API with the use of Artificial Intelligence.

Then, the algorithm Data Mining K-means finds more photos with similar content and organises them in sets, adding them in a multidimensional cloud, and via the algorithm of automatic learning t-Distributed Stochastic Neighbour Embedding (t-SNE) the multidimensional scaling for the data analysis takes place. That is to say, we depict the 1000 cloud dimensions in a human-friendly 3 dimensional form so that anyone can navigate through the results.

It may sound complicated, but this way a concept only known to humans can be encoded in a way that can be recognised and processed by a machine.

So, what is love? Is it a touch? A sunset? A colour? A taste? We will watch the final conclusions together and we will see them gradually getting a shape in this site. Don’t forget to answer and visit again to see how the results develop.