Following are our core areas of interest for potential collaboration:
Homomorphic Encryption (HE) schemes allow computations on ciphertext, without revealing information on plaintext. We design new fully homomorphic encryption (FHE) constructions based on new mathematical tools to efficiently secure data management on the cloud. To improve the efficiency of existing FHE constructions, we develop efficient computation algorithms on encrypted data.
|Symmetric Searchable Encryption (SSE) schemes allow ciphertext searching which enables data to be securely stored at the cloud. Our work focuses on building flexible SSE schemes for ease of configuration in the level of security, search time, storage for application in practical large databases.We design leakage-resilient provable secure frameworks to maximize query capacity and improve existing structured encryption schemes to achieve performance-balanced expressive SSE.
|Multi-Party Computation (MPC) schemes enable multiple parties to collaboratively perform computation without disclosing any party's private input. We design and implement secure MPC based on FHE, speed up the MPC of particular functions in terms of rounds, improve communication and computation complexities.
|Differential Privacy (DP) mechanism can be used to satisfy data aggregation without leaking individual private information. These algorithms rely on incorporating random noise into the data.
|Federated Learning (FL) can be used to locally train a model and broadcast, allowing the learning of predictive models based on data originating from multiple independent data owners without exposing their private data. We aim to tackle the privacy concerns around cross-organizational machine learning and data sharing.