Head of the Department, Computer Science
Professor of Computer Science
Governments around the world – and India in particular – are trying to build large data registries for effective delivery of a variety of public services. However, these efforts are often undermined due to serious concerns over privacy risks associated with collection and processing of personally identifiable information. While a rich set of special- purpose privacy-preserving techniques exist in computer science, they are unable to provide end-to-end protection in alignment with legal principles in the absence of an overarching operational architecture to ensure purpose limitation and protection against insider attacks. This either leads to weak privacy protection in large designs, or adoption of overly defensive strategies to protect privacy by compromising on utility.
We investigate the issues in designing an operational architecture for privacy-by-design based on independent regulatory oversight stipulated by most data protection regimes, regulated access control, purpose limitation and data minimisation.
India’s parliamentary election is the largest in the world, with 543 constituencies and well over 1 million voters per constituency on the average, and voting is conducted electronically since 2004. However, there is considerable doubt about the integrity of both the Electronic Voting Machine (EVM) used by the Election Commission of India (ECI) and the procedure for maintaining and updating voters lists.
We analyse the ECI solutions from the points of view of verifiability and compliance with democratic principles. We also investigate the possibility of end-to-end verifiable electronic voting that can support voter-verified paper audit trails in a closely coupled manner.
Citizens’ Commission on Elections’ Report on EVMs and VVPAT. Madan Lokur, Wajahat Habibullah, Hariparanthaman, Arun Kumar, Subhashis Banerjee, Pamela Philipose, John Dayal, Sundar Burra and M. G. Devasahayam. Economic and Political Weekly, Vol. 57, Issue No. 3, 15 Jan, 2022. (https://www.epw.in/journal/2022/3/perspectives/citizens’-commission-elections’-report-evms-and.html)
In collaboration with the radiology group at AIIMS, New Delhi, we investigate improved detection of breast cancer from mammograms in situations where the presentation is obscure and difficult. We also investigate using other auxiliary information to improve the accuracy of detection.
In a one-off problem, we also addressed detection of Covid from chest X-Ray in a hospital setting.
We have an ongoing collaboration with AIIMS where we study and build simulators for neuroendoscopy skills training.
We develop a set of machine learning based tools for accurate prediction of socio-economic indicators from daytime satellite imagery. The diverse set of indicators are often not intuitively related to observable features in satellite images, and are not even always well correlated with each other. Our predictive tool is more accurate than using night light as a proxy, and can be used to predict missing data, smooth out noise in surveys, monitor development progress of a region, and flag potential anomalies. December 2021. (https://arxiv.org/abs/2202.00109)
We look at large scale (city scale) 3D reconstruction from image collections.
We briefly looked at some problems related to self-driving cars, in particular that of monocular SLAM and ego-motion.
We investigate reliability issues of machine learning, especially with respect to adversarial attacks.
Based on an affidavit in support of a petition filed in the Kerala High Court.
A summary appeared in the Indian Express on May 5, 2018 and was reproduced in Metamorphoses on May 18, 2018.
A related short note with Subodh Sharma appeared in Ideas for India on July 4, 2018.
A shorter version in the Economic and Political Weekly, September 2017.
A summary appeared in the in The Wire, July 2017.
This article received some press coverage, see here.