After a short couple year hiatus, DC Area Crypto Day is Back!!
Date: Friday, May 1st
Time: 10:00-4:00 ET
Location: University of Maryland, Brendan Iribe Center, IRB-5105.
Zoom link: contact kaptchuk at umd dot edu
Parking: Here, for $4 / hour. More information here.
Shuttle: from College Park Metro. Take a left out of the fare gates and climb the stairs. Look for the 104 Shuttle bus (there are signs for it), and exit at the Glenn L. Martin Wind Tunnel stop. Alternatively, the walk is ~20ish minutes
Lunch: We will provide light breakfast and coffee and lunch. If you have particular dietary restrictions, please email kaptchuk at umd dot and we will try our best to accomodate you. The earlier you reach out the better
Schedule (Tentitive):
| Time | Speaker/Event |
| 10:00-10:30 | Welcome/Light breakfast (provided) |
| 10:30-11:30 | Aditya Hegde (JHU) |
| 11:30-12:30 | Natalie Lang (UMD) |
| 12:30-2:00 | Lunch (provided) |
| 2:00-3:00 | Saba Eskandarian (UNC) |
| 3:00-4:00 | Varun Madathil (Yale) |
Speakers:
Aditya Hegde
Title: Client-Server Homomorphic Secret Sharing in the CRS Model
Abstract: Homomorphic secret sharing (HSS) is a distributed analogue of fully homomorphic encryption (FHE) where clients secret-share their inputs among two (or more) servers, who then non-interactively compute additive shares of the function output. Unlike FHE, HSS enables a distributed form of homomorphic computation from a broader set of cryptographic assumptions.
All known multi-client HSS schemes require a correlated setup that is used by the clients to share their inputs. In contrast, multi-client delegation of computation using only a CRS has been known from assumptions that imply FHE since the introduction of multi-key FHE (Lopez-Alt et al., STOC 2012).
We close this gap by providing the first construction of client-server HSS in the CRS only model from DDH, DCR, and more. Our constructions subsume all existing variants of group-based HSS schemes, including public-key HSS (Boyle et al., Crypto 2016), succinct HSS (Abram et al., Eurocrypt 2024), and multi-key HSS (Couteau et al., Eurocrypt 2025), and their applications.
This is joint work with Damiano Abram, Geoffroy Couteau, Lalita Devadas, Abhishek Jain, Lawrence Roy, and Sacha Servan-Schreiber.
Natalie Lang
Title: On the Privacy Aspects of Compression
Abstract: In this talk, I will discuss an unexpected aspect of compression: beyond reducing communication and storage costs, it inherently conceals information and can thereby serve as a tool for enhancing privacy and security. I will illustrate this perspective through two examples from different federated learning settings.
Saba Eskandarian
Title: Abuse Reporting Protocols for Private Messaging
Abstract: Users expect the ability to report abuse as a standard feature on private messaging applications. Unfortunately, deployed verifiable abuse reporting schemes are brittle, relying on platform access to user metadata and having the moderator in-the-loop for message delivery. Solutions that remove these restrictions incur multiple orders of magnitude higher performance costs on both the client and the server.
This talk will first explore how to minimize abuse reporting costs for these challenging settings, showing some settings where we can reduce costs to nearly match deployed schemes, and showing fundamental performance barriers in others.
Next, we will focus on the setting of third party moderation, where the moderator is not in the loop for message delivery. Allowing third parties to serve as moderators broadens the tools available to platforms that cannot (or choose not to) deploy a large in-house moderation apparatus. Here we will demonstrate new schemes that extend platforms’ and users’ policy and enforcement options compared to existing approaches while simultaneously lowering performance costs compared to prior work.
This talk is based on joint work with Matthew Gregoire, Margaret Pierce, Gabriel Schell, and Jade Keegan.
Title: Efficient Secure Aggregation for Federated Learning
Abstract: Federated Learning (FL) trains a global model by having each selected device push only its model update to a central server, keeping raw data local. However, those updates can still leak sensitive information unless the server learns only their sum. A naïve approach is to run a generic secure‑multiparty sum, but off‑the‑shelf protocols require several rounds of interaction and even direct client‑to‑client communication – often infeasible in FL, where mobile devices are intermittently online and can drop out at any moment, and cannot be expected to interact with each other.
In this talk, I will review the secure‑aggregation problem in the context of FL and explain why naïve solutions fail by focusing on constraints unique to the FL setting.
I will then present Tacita, a single‑server protocol that satisfies these FL‑specific constraints while retaining provable security. Tacita uses an external committee (needed to prevent residual leakage) to aid in secure aggregation and offers:
- One‑shot execution: every client and every committee member sends exactly one message.
- Constant‑size communication per client, independent of the round’s cohort of clients or committee size.
- Robust aggregation despite client or committee dropout.
These properties are enabled by two new primitives: (i) succinct multi‑key linearly homomorphic threshold signatures (MKLHTS) for verifiable input soundness with a single aggregate signature, and (ii) a homomorphic variant of Silent Threshold Encryption (CRYPTO ’24).
DC area Crypto Day 
