MIT Department: Media Arts and Sciences
Undergraduate Institution: University of Florida
Faculty Mentor: Sandy Pentland
Research Supervisor: Thomas Hardjono, Christian Smith
I am from Plantation, Florida and I am currently studying Computer Science at University of Florida. My research interests are in the fields of human computer interaction and internet of things. My hobbies include reading, scuba diving, road tripping to new places, and listening to music. I hope to obtain a PhD after my undergraduate studies and create my own business thereafter relating to IoT.
2017 Research Abstract
Using Smart Algorithms for Better Twitter Privacy and for Fitbit Computations
Eric Agredo, Department of Computer & Information Science & Engineering, University of Florida
Thomas Hardjono and Alex ‘Sandy’ Pentland, Human Dynamics, Department of Media Arts and Sciences, Massachusetts Institute of Technology
With an exponential growth of our technology comes an even larger growth in our data. Data is omnipresent in our lives through sensors, our mobile phones, social network, financial devices, and wearable devices. Since data has become an important part of our lives, it is essential that we implement protocols to ensure a right to ownership of our data and a secure manner to share data. The Human Dynamics group has created the Open Algorithm(OPAL) protocol to do exactly this. We began to use OPAL on notoriously public data such as Twitter. We looked at the location and topic of every tweet within the greater Boston area in order to target certain areas of Boston based on different to pics. We then implemented this with the OPAL protocol to demonstrate to Twitter that we could keep a user’s twitter information completely private while preserving the value of a query response. Then, we used a user’s Fitbit data as the data provider. Once a Fitibit user uploads their data through the OPAL protocol, a user can obtain useful query responses on their own data while maintaining privacy on it. As of now, Twitter gives raw data to any user it pleases and as a result malicious queries can be run on the data. Also, Fitbit data is extremely valuable, but doctors are not knowledgeable on how to use it. Our project shows that computations can be run on propriety data to receive responses while maintaining 100% privacy.