Twitter
is indeed on a push to become more like Facebook, it seems.
This
as the musings of Twitter CEO Dick Costolo at their recent Earning’s Call as
reported in “Twitter
may be considering a Facebook-style feed — but would that help its growth or
derail it?”, published JUL. 30, 2014 - 4:13 PM PDT by Mathew Ingram, GigaOM now has some real-life evidence to back
them up!
This
as they’ve purchased MadBits, a year old New
York-based image Search company for an undisclosed sum as reported in “Twitter
acquires Madbits for 'deep learning' skills”, published July 30, 2014 9:52
AM PDT by Charlie Osborne, CNET News and “Twitter
acquires deep learning startup Madbits”, published JUL. 29, 2014 - 4:58 PM
PDT by Derrick Harris, GigaOM.
MadBits is a Deep Learning Platform that the
co-founding pair of Clement Farabet and Louis-Alexandre Etezad-Heydari have
trained to recognize pictures as pointed out in “Twitter
joins deep learning trend, acquires Madbits”, published July 30, 2014 9:25
AM, VentureBeat. In essence, they wrote
an AI (Artificially Intelligent) program and then taught it the rules of how to
recognize traits in pictures.
While
Louis-Alexandre Etezad-Heydari worked as a Senior Software Engineer at Twitter,
fellow co-founder Clément Farabet, was a research scientist at Courant Institute at New York University.
Curiously, both interned under Facebook AI Lab Director Yann LeCun, who
happened to be the PhD Supervisor for Clément Farabet. So, clearly MadBits has clever scientists with backgrounds
in AI development, a natural fit to Twitter.
Twitter buys Madbits –
How Torch7 that powers Artificial Vision makes Twitter more like Facebook
MadBits uses Torch7, a Fourth Generation AI
Programming platform based on the scripting language LuaJIT with a dash of C Language. Torch7, which is freely available on Github,
was accessed by Clement Farabet, who according to his personal website,
was deep into researching Artificial Vision.
Together,
the pair worked on getting this tool to makes sense of the world we humans and
animals do and learn from what they see. As machines can't really make anything
of images, they have to teach it some basic and advanced rules, making MadBits, the outcome of their efforts, a company
that found practical application for data mining from images.
Somehow
this doesn’t sound good, as I’d predicted this very same thing after Twitter
had introduce Tagging and multi-photo posting as explained in my Geezam blog article entitled “Twitter
add Photo-Tagging and uploading Multiple Photos as Mobile Social Media Evolves”.
After
all, those are Facebook-esque traits, similar to their Facebook makeover back
in April 2014 as detailed in my blog article entitled
“How
to Download Twitter user Pictures as Twitter makes Facebook Timeline changes”.
Guess their Earnings Call results prompted them to take actions to appease
shareholders as it relates to growing the company.
So
if I’m reading the Tea Leave and my Tarot Cards right, we might be seeing a
Twitter Graph in the Future, an analogous counterweight to Facebook Graph
Search as described in my Geezam blog
article entitled “Facebook
to introduce #Hashtags with emoticons, monetizing FB and making Open Graph
Search Practical”, but geared towards pictures.
Stay
tuned to my blog for more Details on this Big Brother Development.
No comments:
Post a Comment