I am an assistant professor of Electrical and Computer Engineering at Carnegie Mellon University. I have a courtesy appointment in the Computer Science Department and am a part of CyLab. My research interests span the algorithmic foundations of blockchains, machine learning, distributed systems, and privacy-preserving technologies.
9/15/2018: I am honored to receive the 2018-2019 World Economic Forum Global Future Council Fellowship in Cybersecurity.
8/25/2018: Alankar Jain has been named a 2018-2019 Siebel Scholar. Congratulations, Alankar!
8/15/2018: Our work on anonymous routing for cryptocurrencies (Dandelion) has been assigned a Bitcoin Improvement Proposal (BIP) number (156). It also has been/is being integrated into BitMessage, Grin, and soon ZCoin.
3/26/2018: Check out Zinan Lin's and Sewoong Oh's YouTube interview on our PacGAN paper.
10/2/2017: Zinan Lin was named a CMU Presidential Fellow for the 2017-2018 academic year. Congratulations, Zinan!
Blockchains are useful for storing data in distributed systems with limited trust. I am interested in designing scalable blockchains that account for resource constraints in the network and in individual devices. This work ranges from protecting users' privacy to building faster consensus algorithms. A common theme in this work relies on explicitly modeling device or network behavior, and using these models to design more efficient algorithms with theoretical guarantees.
Generative adversarial networks (GANs) are a technique for learning a generative model from data. They have been tremendously successful at producing high-quality, sharp images. However, they are not well-understood. I am interested in studying the dynamics of GANs themselves (e.g. improving diversity and interpretability), as well as using them for the release of privacy-preserving datasets.
Recent years have brought increasing levels of surveillance. I am interested in designing privacy-preserving algorithms that enable people to communicate freely without sacrificing privacy. I have been working on a few main problems within this theme, related to anonymous social media (e.g., Yik Yak, Secret) and anonymous peer-to-peer networks (e.g., Bitcoin, cryptocurrencies). A common theme in this work is that we wish to provide statistical anonymity guarantees against computationally-unbounded adversaries.
gfanti (at) andrew (dot) cmu (dot) edu
2118 Collaborative Innovation Center
Carnegie Mellon University
4720 Forbes Ave, Pittsburgh, PA 15213