Property Inference from Poisoning
Explore techniques for inferring properties from poisoned data, examining implications for privacy and security in machine learning systems.
SNARKBlock - Federated Anonymous Blocklisting from Hidden Common Input Aggregate Proofs
Explore federated anonymous blocklisting using hidden common input aggregate proofs, enhancing privacy and security in distributed systems.
Surakav - Generating Realistic Traces for a Strong Website Fingerprinting Defense
Explore a novel defense against website fingerprinting attacks using Surakav, a technique for generating realistic network traces to enhance online privacy and security.
Augury - Using Data Memory-Dependent Prefetchers to Leak Data at Rest
Explore a novel data leakage technique exploiting memory-dependent prefetchers, revealing vulnerabilities in modern processors and implications for cybersecurity.
Naos - Serialization-free RDMA Networking in Java
Innovative technique for direct object transfer between Java heaps over RDMA networks, bypassing serialization for improved performance in distributed systems.
Sustainable Software Engineering and SREs
Explore how SREs can champion technical, operational, and environmental sustainability efforts, balancing reliability with eco-friendly practices in software engineering.
Generalized Sub-Query Fusion for Eliminating Redundant I-O from Big-Data Queries
Explores RESIN optimizer extension for SQL queries, introducing Generalized Sub-Query Fusion to eliminate redundant I/O and improve performance in big-data analytics, with significant speedups demonstrated on TPCDS benchmark.
A Tensor Compiler Approach for One-size-fits-all ML Prediction Serving
Innovative tensor compiler approach for efficient ML model scoring, simplifying infrastructure and leveraging neural network compilers for CPU and GPU acceleration.
Virtual Consensus in Delos
Explores virtualizing consensus in replicated systems, enabling dynamic protocol changes without downtime. Introduces VirtualLog and Loglets for simplified, diverse, and composable consensus implementations, showcasing benefits in the Delos database.
Rammer - Enabling Holistic Deep Learning Compiler Optimizations with rTasks
Optimizing DNN computation on hardware accelerators through novel abstractions and holistic scheduling, achieving significant performance improvements over existing compilers and libraries.
Testing Database Engines via Pivoted Query Synthesis
Innovative approach to detect logic bugs in database systems using Pivoted Query Synthesis, demonstrating high effectiveness in finding and fixing issues in major DBMSs.
Trojan Source and Bad Characters - Invisible Hacks and Reluctant Patching
Explore invisible hacks and reluctant patching in software security, focusing on Trojan Source attacks and the challenges of addressing hidden vulnerabilities in code.
TROLLMAGNIFIER - Detecting State-Sponsored Troll Accounts on Reddit
Explore techniques for identifying state-sponsored troll accounts on Reddit, examining detection methods, challenges, and implications for online discourse and platform integrity.
Desperate Times Call for Desperate Measures - User Concerns with Mobile Loan Apps in Kenya
Explore user concerns and challenges with mobile loan apps in Kenya, examining their impact on financial practices and potential risks for vulnerable populations.
Security Foundations for Application-Based Covert Communication Channels
Explore covert communication channels in applications, focusing on security foundations, techniques, and potential implications for network security and privacy.
Sphinx - Enabling Privacy-Preserving Online Learning over the Cloud
Explore privacy-preserving techniques for online learning in cloud environments, focusing on data protection and secure computation methods.
Membership Inference Attacks from First Principles
Explore the fundamentals of membership inference attacks, their implications, and defense strategies in this comprehensive IEEE talk by leading security researchers.
BadEncoder - Backdoor Attacks to Pre-trained Encoders in Self-Supervised Learning
Explore backdoor attacks on pre-trained encoders in self-supervised learning, examining vulnerabilities and potential security risks in AI systems.
Bad Characters - Imperceptible NLP Attacks
Explore imperceptible attacks on natural language processing systems, examining vulnerabilities and potential defenses in AI-driven text analysis.
Flawed, but Like Democracy We Don't Have a Better System - The Experts' Insights on the Peer Review Process of Evaluating Security Papers
Experts discuss insights on the peer review process for security papers, exploring flaws and potential improvements in academic evaluation.
Privacy-from-Birth - Protecting Sensed Data from Malicious Sensors with VERSA
Explore VERSA, a novel approach to safeguarding sensed data from malicious sensors, ensuring privacy protection from the moment of data collection in IoT and cyber-physical systems.