Demystifying AI was the title of my session at Mozfest 2016. In this blog post, I plan to write about what I had discussed with my audience in my fireside session.
Check out my poster which I made for my session https://twitter.com/dvigneshwer/status/792039340784381952
Well I am not really good at poster making never has been my cup of tea but I had compensated it with my jupyter notebook on the classification of male or female ( https://twitter.com/dvigneshwer/status/790215686006054912).
Artificial intelligence is a very big topic and has been a topic of discussion for a very long time, man has always imagined computers to do tasks and jobs in human level efficiency.
Well, one of my sessions primary agenda was to inform my audience that era of AI is near and we have already started being a passive part of it in our day to day life.
I had got 30mins to discuss the things I had put on the poster and each topic had a story which correlates to my two years of research in this field.
I started by asking what is AI ? to my audience, well people were very silent they were all in an impression that AI is like some supernatural things. I won’t complain about that fact like the movies having been picturing AI like that.
I started off with what do you people feel about JARVIS the controller of IRON man suit, after which I explained brain model for AI which is connected to various parts like eyes (to see) , ears (to listen) and mouth (to speak).
Recent efforts of research analysts have been to achieve human level efficiency in each of this parts and complex mathematical model to complement its performance.
I had been working on eyes part of it which is the field of computer vision. Some papers which I have tried to implement is the Alexnet, Lenet, Google net and Deepface. These paper are behind the technologies of facebook image tagging and google image apps.
Later in the session I talked about the different smart apps in their phones and websites we are using, basically all these algorithms are trained to achieve human level efficiency.
With all the diverse data available now due to the technologies and device available is mind blowing. These data will help researchers fine-tune their models to achieve a greater result.
I started questioning my session attendees “Do you guys really know what is happening with our data and the way we interact with the AI web products?” .
Seriously there is great need of companies to come out and explain what people have been doing with our data and our properties.
Things like Skynet etc will never exist but we need to fear for the fact that few companies have a very major share of the AI industry and this is really scary as days go on these decision systems becomes a part of our life and we need to make sure that things are decentralized and power is in the hand of the user.
At the last section of my session was a demo in which I made a classifier differentiate between male and female. I shared the algorithm details and libraries which will help individuals get started.
Github repo which contains the content of my presentation https://github.com/dvigneshwer/DemystifyingAI/
Slides @ http://slides.com/vigneshwerdhinakaran/deck-8#/
Get more relevant links and fireside chat notes at https://public.etherpad-mozilla.org/p/mozfest-2016-demystifying-artificial-intelligence
Find the complete details about what was cooking at Open science space @https://science.mozilla.org/blog/2016-mozfest-recap