Last December, we had an opportunity to attend MSR winter school at IIIT-H. It was an awesome experience, got a chance to meet Prof. Sanjeev Arora, HOD of Princeton's TCS department, Prof. C. Pandurangan, PHD mentor of Kannan sir, and a leading authority in digital signatures. So far, the title and introduction has no correlation, and that is where Cynthia Dwork comes. She is a leading researcher at M.S.R and is working on differential privacy.
The theory of Differential Privacy works on measures to protect sensitive information from public records. Well, in order to understand the working, we have pretty interesting cases, but I will focus my efforts on this new trend of our college called IiiT-h compliments. Since the community can't be accessed presently, I will summarize a few facts. It uses anonymous expression of opinions or thoughts directed to some particular individual. This anonymity is supported by a google forms.
Now the claim is that, this anonymity is just a variable under-processed. When you bring in more public databases, in this particular case, facebook likes, acute observation at public places like cafeteria, canteen etc, or rumours, or information that only your friends know; my argument is that, using these information, you can unveil this cloak of anonymity.
I tried out this system, and put an info that only I and my wing-mates knew, and I thought, it will be pretty obvious to the guy whom I was directing to. Now, knowing my eccentric moods, and the fact that I was the one who uncovered that information, the first door knocked will be mine. That got me thinking, is it really anonymous.
Here the anonymity fails because, there are too many loose ends, and bit by bit, it will eventually become a byte. So, all we need a Sherlock and zeal to connect the dots.
The theory of Differential Privacy works on measures to protect sensitive information from public records. Well, in order to understand the working, we have pretty interesting cases, but I will focus my efforts on this new trend of our college called IiiT-h compliments. Since the community can't be accessed presently, I will summarize a few facts. It uses anonymous expression of opinions or thoughts directed to some particular individual. This anonymity is supported by a google forms.
Now the claim is that, this anonymity is just a variable under-processed. When you bring in more public databases, in this particular case, facebook likes, acute observation at public places like cafeteria, canteen etc, or rumours, or information that only your friends know; my argument is that, using these information, you can unveil this cloak of anonymity.
I tried out this system, and put an info that only I and my wing-mates knew, and I thought, it will be pretty obvious to the guy whom I was directing to. Now, knowing my eccentric moods, and the fact that I was the one who uncovered that information, the first door knocked will be mine. That got me thinking, is it really anonymous.
Here the anonymity fails because, there are too many loose ends, and bit by bit, it will eventually become a byte. So, all we need a Sherlock and zeal to connect the dots.
Image taken from google