Explanipedia

All Rivers Run to the Sea: Private Learning with Asymmetric Flows Open

Yue Niu, Ramy E. Ali, Saurav Prakash, Salman Avestimehr · 2023

Data privacy is of great concern in cloud machine-learning service platforms, when sensitive data are exposed to service providers. While private computing environments (e.g., secure enclaves), and cryptographic approaches (e.g., homomorph…

Securing Secure Aggregation: Mitigating Multi-Round Privacy Leakage in Federated Learning Open

Jinhyun So, Ramy E. Ali, Başak Güler, Jiantao Jiao, A. Salman Avestimehr · 2023

Secure aggregation is a critical component in federated learning (FL), which enables the server to learn the aggregate model of the users without observing their local models. Conventionally, secure aggregation algorithms focus only on ens…

Comments on CausalEC: A Causally Consistent Data Storage Algorithm Based on Cross-Object Erasure Coding Open

Ramy E. Ali · 2023

Cadambe and Lyu 2021 presents an erasure coding based algorithm called CausalEC that ensures causal consistency based on cross-object erasure coding. This note shows that the algorithm presented in Cadambe and Lyu 2021 and the main ideas b…

3LegRace: Privacy-Preserving DNN Training over TEEs and GPUs Open

Yue Niu, Ramy E. Ali, Salman Avestimehr · 2022

Leveraging parallel hardware (e.g. GPUs) for deep neural network (DNN) training brings high computing performance. However, it raises data privacy concerns as GPUs lack a trusted environment to protect the data. Trusted execution environme…

ApproxIFER: A Model-Agnostic Approach to Resilient and Robust Prediction Serving Systems Open

Mahdi Soleymani, Ramy E. Ali, Hessam Mahdavifar, A. Salman Avestimehr · 2022

Due to the surge of cloud-assisted AI services, the problem of designing resilient prediction serving systems that can effectively cope with stragglers and minimize response delays has attracted much interest. The common approach for tackl…

Adaptive Verifiable Coded Computing: Towards Fast, Secure and Private Distributed Machine Learning Open

Tingting Tang, Ramy E. Ali, Hanieh Hashemi, Tynan Gangwani, Salman Avestimehr , et al. · 2022

Stragglers, Byzantine workers, and data privacy are the main bottlenecks in distributed cloud computing. Some prior works proposed coded computing strategies to jointly address all three challenges. They require either a large number of wo…

Secure Aggregation for Buffered Asynchronous Federated Learning Open

Jinhyun So, Ramy E. Ali, Başak Güler, Amir Salman Avestimehr · 2021

Federated learning (FL) typically relies on synchronous training, which is slow due to stragglers. While asynchronous training handles stragglers efficiently, it does not ensure privacy due to the incompatibility with the secure aggregatio…

AsymML: An Asymmetric Decomposition Framework for Privacy-Preserving DNN Training and Inference Open

Yue Niu, Ramy E. Ali, Salman Avestimehr · 2021

Leveraging parallel hardware (e.g. GPUs) to conduct deep neural network (DNN) training/inference, though significantly speeds up the computations, raises several data privacy concerns. Trusted execution environments (TEEs) have emerged as …

3LegRace: Privacy-Preserving DNN Training over TEEs and GPUs Open

Yue Niu, Ramy E. Ali, Salman Avestimehr · 2021

Leveraging parallel hardware (e.g. GPUs) for deep neural network (DNN) training brings high computing performance. However, it raises data privacy concerns as GPUs lack a trusted environment to protect the data. Trusted execution environme…

LightSecAgg: a Lightweight and Versatile Design for Secure Aggregation in Federated Learning Open

Jinhyun So, Chaoyang He, Chien-Sheng Yang, Songze Li, Qian Yu , et al. · 2021

Secure model aggregation is a key component of federated learning (FL) that aims at protecting the privacy of each user's individual model while allowing for their global aggregation. It can be applied to any aggregation-based FL approach …

ApproxIFER: A Model-Agnostic Approach to Resilient and Robust Prediction Serving Systems Open

Mahdi Soleymani, Ramy E. Ali, Hessam Mahdavifar, A. Salman Avestimehr · 2021

Due to the surge of cloud-assisted AI services, the problem of designing resilient prediction serving systems that can effectively cope with stragglers/failures and minimize response delays has attracted much interest. The common approach …

Verifiable Coded Computing: Towards Fast, Secure and Private Distributed Machine Learning. Open

Tingting Tang, Ramy E. Ali, Hanieh Hashemi, Tynan Gangwani, Salman Avestimehr , et al. · 2021

Stragglers, Byzantine workers, and data privacy are the main bottlenecks in distributed cloud computing. Several prior works proposed coded computing strategies to jointly address all three challenges. They require either a large number of…

Adaptive Verifiable Coded Computing: Towards Fast, Secure and Private Distributed Machine Learning Open

Tingting Tang, Ramy E. Ali, Hanieh Hashemi, Tynan Gangwani, Salman Avestimehr , et al. · 2021

Stragglers, Byzantine workers, and data privacy are the main bottlenecks in distributed cloud computing. Some prior works proposed coded computing strategies to jointly address all three challenges. They require either a large number of wo…

List-Decodable Coded Computing: Breaking the Adversarial Toleration Barrier Open

Mahdi Soleymani, Ramy E. Ali, Hessam Mahdavifar, A. Salman Avestimehr · 2021

We consider the problem of coded computing, where a computational task is performed in a distributed fashion in the presence of adversarial workers. We propose techniques to break the adversarial toleration threshold barrier previously kno…

Securing Secure Aggregation: Mitigating Multi-Round Privacy Leakage in Federated Learning Open

Jinhyun So, Ramy E. Ali, Başak Güler, Jiantao Jiao, Salman Avestimehr · 2021

Secure aggregation is a critical component in federated learning (FL), which enables the server to learn the aggregate model of the users without observing their local models. Conventionally, secure aggregation algorithms focus only on ens…

Consistency Analysis of Replication-Based Probabilistic Key-Value Stores Open

Ramy E. Ali · 2021

Partial quorum systems are widely used in distributed key-value stores due to their latency benefits at the expense of providing weaker consistency guarantees. The probabilistically bounded staleness framework (PBS) studied the latency-con…

On Polynomial Approximations for Privacy-Preserving and Verifiable ReLU Networks Open

Ramy E. Ali, Jinhyun So, Amir Salman Avestimehr · 2020

Outsourcing deep neural networks (DNNs) inference tasks to an untrusted cloud raises data privacy and integrity concerns. While there are many techniques to ensure privacy and integrity for polynomial-based computations, DNNs involve non-p…

Hierarchical Deep Double Q-Routing Open

Ramy E. Ali, Bilgehan Erman, Ejder Baştuǧ, Bruce Cilli · 2020

This paper explores a deep reinforcement learning approach applied to the packet routing problem with high-dimensional constraints instigated by dynamic and autonomous communication networks. Our approach is motivated by the fact that cent…

Fundamental Limits of Erasure-Coded Key-Value Stores With Side Information Open

Ramy E. Ali, Viveck R. Cadambe, Jaime Llorca, Antonia M. Tulino · 2020

In applications of distributed storage systems to modern key-value stores, the stored data is highly dynamic due to frequent updates. The multi-version coding problem was formulated to study the cost of storing dynamic data in distributed …

Info-Commit: Information-Theoretic Polynomial Commitment Open

Saeid Sahraei, Salman Avestimehr, Ramy E. Ali · 2020

We introduce Info-Commit, an information-theoretic protocol for polynomial commitment and verification. With the help of a trusted initializer, a succinct commitment to a private polynomial $f$ is provided to the user. The user then querie…

Harnessing Correlations in Distributed Erasure-Coded Key-Value Stores Open

Ramy E. Ali, Viveck R. Cadambe · 2019

Motivated by applications of distributed storage systems to key-value stores, the multi-version coding problem was formulated to efficiently store frequently updated data in asynchronous decentralized storage systems. Inspired by consisten…

Fundamental Limits of Erasure-Coded Key-Value Stores with Side\n Information Open

Ramy E. Ali, Viveck R. Cadambe, Jaime Llorca, Antonia M. Tulino · 2018

In applications of distributed storage systems to modern key-value stores,\nthe stored data is highly dynamic due to frequent updates. The multi-version\ncoding problem was formulated to study the cost of storing dynamic data in\ndistribut…

Multi-version Coding with Side Information Open

Ramy E. Ali, Viveck R. Cadambe, Jaime Llorca, Antonia M. Tulino · 2018

In applications of storage systems to modern key-value stores, the stored data is highly dynamic due to frequent updates from the system write clients. The multi-version coding problem has been formulated to study the cost of storing dynam…

Multi-version Coding for Consistent Distributed Storage of Correlated Data Updates. Open

Ramy E. Ali, Viveck R. Cadambe · 2017

Ramy E. Ali YOU? Author Swipe