Publications by Rachid Guerraoui

Chat with Graph Search

Ask any question about EPFL courses, lectures, exercises, research, news, etc. or try the example questions below.

DISCLAIMER: The Graph Chatbot is not programmed to provide explicit or categorical answers to your questions. Rather, it transforms your questions into API requests that are distributed across the various IT services officially administered by EPFL. Its purpose is solely to collect and recommend relevant references to content that you can explore to help you answer your questions.

Federated learning (FL) is an appealing approach to training machine learning models without sharing raw data. However, standard FL algorithms are iterative and thus induce a significant communication cost. One-shot federated learning (OFL) trades the iter ...

An indulgent algorithm is a distributed algorithm that, besides tolerating process failures, also tolerates arbitrarily long periods of instability, with an unbounded number of timing and scheduling failures. In particular, no process can take any irrevoca ...

Software transactional memories (STM) are described in the literature with assumptions of sequentially consistent program execution and atomicity of high level operations like read, write, and abort. However, in a realistic setting, processors use relaxed ...

Amodernmanycore architecture can be viewed as a distributed system with explicit message passing to communicate between cores. Ensuring the consistency of shared state replicated over several cores is the key to the well functioning of such a system. Yet, ...

While spreading fake news is an old phenomenon, today social media enables misinformation to instantaneously reach millions of people. Content-based approaches to detect fake news, typically based on automatic text checking, are limited. It is indeed diffi ...

Consensus and State Machine Replication (SMR) are generally considered to be equivalent problems. In certain system models, indeed, the two problems are computationally equivalent: any solution to the former problem leads to a solution to the latter, and v ...

This paper addresses the problem of synchronous beeping, as addressed by swarms of fireflies. We present Byzantine-resilient algorithms ensuring that the correct processes eventually beep synchronously despite a subset of nodes beeping asynchronously. We assume that n > 2f (n is the number of processes and f is the number of Byzantine processes) and that the initial state of the processes can be arbitrary (self-stabilization). We distinguish the cases where the beeping period is known, unknown or approximat

This paper presents a transactional framework for low-latency, high-performance, concurrent event processing in Java. At the heart of our framework lies Reﬂexes, a restricted programming model for highly responsive systems. A Reﬂex task is an event process ...

Event dissemination in large scale dynamic systems is typically claimed to be best achieved using decentralized peer-to-peer architectures. The rationale is to have every participant in the system act both as a client (information consumer) and as a server ...

Pseudo-code descriptions of STMs assume sequentially consistent program execution and atomicity of high-level STM operations like read, write, and commit. These assumptions are often violated in realistic settings, as STM implementations run on relaxed mem ...

We propose uBFT, the first State Machine Replication (SMR) system to achieve microsecond-scale latency in data centers, while using only 2f+1 replicas to tolerate f Byzantine failures. The Byzantine Fault Tolerance (BFT) provided by uBFT is essential as pu ...

We present TRIAD, a new persistent key-value (KV) store based on Log-Structured Merge (LSM) trees. TRIAD improves LSM KV throughput by reducing the write amplification arising in the maintenance of the LSM tree structure. Although occurring in the backgrou ...

Despite their distributed nature, blockchains frequently experience failures that disrupt availability, delay transactions, and, in some cases, cause total network outages. Ensuring fault tolerance is critical for the reliability of blockchain-based applic ...

This paper addresses the problem of determining the weakest failure detector to implement consensus in a message passing system when t out of n processes can crash (including when n/2 =< t < n-1), by addressing the problem of determining the weakest failure detector to implement a register. We complement and, in a precise sense, generalise previous results on the implementability of consensus and registers in a message passing model (augmented with the failure detector abstraction).

Recommenders personalize the web content using collaborative filtering to relate users (or items). This work proposes to unify user-based, item-based and neural word embeddings types of recommenders under a single abstraction for their input, we name Consu ...

The power of a set S of object types can be measured as the maximum number n of processes that can solve consensus using only types in S and registers. This number, denoted by h^r_m (S), is called the consensus power of S. The use of failure detectors can however ``boost'' the consensus power of types. This paper addresses the weakest failure detector type booster question, which consists in determining the weakest failure detector D such that, for any set S of types with h_m^r(S)=n, h_m^r(S;D)=n+1. We cons

Modern Byzantine fault-tolerant state machine replication (BFT) protocols involve about 20,000 lines of challenging C++ code encompassing synchronization, networking and cryptography. They are notoriously difficult to develop, test and prove. We present a ...

Renaming is a fundamental problem in distributed computing, in which a set of n processes need to pick unique names from a namespace of limited size. In this paper, we present the first early-deciding upper bounds for synchronous renaming, in which the run ...

Rachid Guerraoui

Chat with Graph Search

Revisiting Ensembling in One-Shot Federated Learning

A General Characterization of Indulgence (Invited Paper)

Verification of STM on relaxed memory models

PaxosInside

The Fake News Vaccine

Special issue on NETYS selected papers

State Machine Replication is More Expensive than Consensus

Byzantine Fireflies

Byzantine Fireflies

High-Performance Transactional Event Processing

Towards Fair Event Dissemination

Software Transactional Memory on Relaxed Memory Models

uBFT: Microsecond-Scale BFT using Disaggregated Memory

TRIAD: creating synergies between memory, disk and log in log structured key-value stores

Evaluating Blockchain Fault Tolerance with Stabl

Failure Detection Lower Bounds on Registers and Consensus (Preliminary Version)

Unified and Scalable Incremental Recommenders with Consumed Item Packs

On Failure Detectors and Type Boosters

The Next 700 BFT Protocols

Early Deciding Synchronous Renaming in O( logf ) Rounds or Less