- Privacy-Preserving Computing
In several scenarios, there is a need to match a query against a dataset, where the query/dataset belongs to different parties and each of them requires keeping their own data private. The importance of this requirement arises in many various areas, e.g., medical history and criminal data. A frequent application of privacy-preserving scenario is matching. For example, Alice wants to find if she has a genetic disorder by matching her genome information with Bob’s genetic disorder bank. But she doesn’t want to reveal her private information and so does Bob.This project aims to address the privacy-preserving matching using Yao’s Garbled Circuit (GC) protocol. GC protocol has shown to be the most efficient secure two party computation approach. This protocol allows two parties to evaluate a function which is described as a Boolean circuit on their private data. This project objective is to study the applicability of GC-based privacy preserving protocols on real benchmarks and optimize its performance for real application on embedded or reconfigurable devices.
- Fast K-Nearest Neighbor Search (KNN)
During fall 2014, I was working on developing a fast new KNN search and optimize it in order to implement it on real hardware platform.
Presentation on reviewing state-of-the-art approach (FLANN)
- SSVEP (Steady State Visual Evoked Potential) based BCI (Brain-Computer Interface)
This research was my B.Sc. project and the aim was to create a SSVEP-based Brain Computer Interface (BCI) for the application of Intelligent Phone-Dialing. You can see the details in Academic Projects section.
“Large-Scale Privacy-Preserving Matching and Search”, August 2016
- M S Riazi, C Weinert, O Tkachenko, E M Songhori, T Schneider, and F Koushanfar. Chameleon: A Hybrid Secure Computation Framework for Machine Learning Applications. Cryptology ePrint Archive: Report 2017/1164
- E M Songhori, M S Riazi, S U Hussain, A R Sadeghi, and F Koushanfar. ARM2GC: Simple and Efficient Garbled Circuit Framework by Skipping. Cryptology ePrint Archive: Report 2017/1157
- M S Riazi, M Samragh, F Koushanfar. “CAMsure: Secure Content-Addressable Memory for Approximate Search”. ACM Transactions on Embedded Computing Systems (TECS), 2017.
* Best paper award nominee
- B D Rouhani, M S Riazi, and F Koushanfar. “DeepSecure: Scalable Provably-Secure Deep Learning”. arXiv preprint arXiv:1705.08963 (2017)
- M S Riazi, E M Songhori, and F Koushanfar. “PriSearch: Efficient Search on Private Data”, Design Automation Conference (DAC) 2017
- M S Riazi, B Chen, A Shrivastava, D Wallach, and F Koushanfar. “Sub-linear Privacy-preserving Search with Untrusted Server and Semi-honest Parties”, arXiv preprint arXiv:1612.01835
- M S Riazi, EM Songhori, AR Sadeghi, T Schneider, and F Koushanfar. “Toward Practical Secure Stable Matching”, Proceedings on Privacy Enhancing Technologies (PoPETs) 2017 **
- M S Riazi, NKR Dantu, LNV Gattu, and F Koushanfar. “GenMatch: Secure DNA Compatibility Testing”, Hardware Oriented Security and Trust (HOST) 2016
** The Early Termination Technique (ETT) from our “Toward Practical Secure Stable Matching” paper has been used in the secure execution of Calctopia which provides secure spreadsheets (https://www.calctopia.com/security-model/).
(Images from calctopia.com)