LACUNA is a new testbed that injects PII into predefined model parameters to benchmark the localization precision of LLM unlearning methods, revealing that SOTA approaches are imprecise despite strong output performance.
hub Canonical reference
ARBITRAR: user-guided API misuse detection
Canonical reference. 100% of citing Pith papers cite this work as background.
45
Pith papers citing it
Background
100% of classified citations
hub tools
citation-role summary
background 9
citation-polarity summary
roles
background 8polarities
background 8representative citing papers
k-REWB matching cannot be solved in O(n to the 2k minus epsilon) time under SETH, is W[2]-hard parameterized by expression length, and 2-use 2-REWBs require superlinear time unless triangle detection does; 1-use REWBs admit an O(n log squared n) algorithm.
A general framework defends any shuffle-DP protocol against poisoning attacks on union-preserving queries, retaining asymptotically equivalent error without attacks and only polylogarithmic increase with constant attackers.
SuperDP refutes ε-DP via simultaneous synthesis of input pairs and witness functions using upper expectation supermartingales and lower expectation submartingales, delivering the first fully automated, sound, and semi-complete method applicable to both discrete and continuous stochastic mechanisms.
A game semantics framework delivers the first sound and complete open-world assertion checking for Ethereum smart contracts, implemented as YulTracer which achieves perfect recall and precision on reentrancy benchmarks and real-world exploits.
CrypFormBench is a new benchmark jointly covering symbolic and computational security to evaluate LLMs on five formal analysis capabilities, with results showing top model Claude-3.5 scores 48.7/100 and most models struggling on generation, transformation, and correction.
VIPIR introduces two new PIR protocols, ExpPack compression, and GPU optimizations for NTT and GEMM that deliver orders-of-magnitude higher throughput than prior systems.
ATWU jointly optimizes model parameters and token weights via a linear scorer on hidden states, recovering oracle forget-specific tokens under a separation condition and achieving SOTA forget-retain trade-offs on TOFU and RWKU.
Introduces bPk# as a delegatable pseudonym system with a formal framework, generic construction with security proofs, concrete instantiation, and reference implementation for eID applications.
A Bayesian decision framework with probabilistic programs computes membership posteriors for Bayesian network populations and outperforms marginal-based attacks on complex dependency structures.
Introduces Unlearning Depth Score (UDS) via activation patching to quantify LLM unlearning depth and claims it outperforms 20 other metrics in faithfulness and robustness on 150 models.
OverrideFuzz uses semantic-aware grammar fuzzing with reflection to model override hooks and dynamic rebinding, producing coverage growth and inputs that match known vulnerability patterns on CPython, Lua, and QuickJS without discovering new bugs in the evaluation window.
A low-stake adversary can degrade a liquid staking pool's performance via consensus manipulation and profit from the resulting drop in its LST value through application-layer financial positions.
Constructs a logical-relations security model for where-declassification in higher-order languages by halting indistinguishability enforcement after relevant declassifications, yielding stronger guarantees than prior lower-order definitions.
A taxonomy of GitHub abuse behaviors is proposed along with a detection framework achieving F1-scores exceeding 89% on a manually labeled dataset of 392 instances.
Privatar uses horizontal frequency partitioning and distribution-aware minimal perturbation to enable private offloading of VR avatar reconstruction, supporting 2.37x more users with modest overhead.
Semantic typing via coinductively defined interpretations on a typed operational semantics ensures information flow control and non-interference for TinySol contracts that use fallback functions.
Persistent BitTorrent Trackers use smart contracts and aggregated cryptographic attestations to make reputation verifiable and resilient to tracker shutdowns, with a hybrid signature scheme and authenticated DHT fallback for decentralized access control.
Introduces the first compositional game-theoretic framework for blockchains using cross-layer games and cross-application composition to analyze incentive compatibility and reveal vulnerabilities in composed systems.
NESA presents a neuro-symbolic framework that decomposes static analyses into policy-defined sub-problems solved by parsers and LLMs to enable compilation-free customizable analysis with reduced hallucinations.
Flutter achieves 2Δ + ε good-case latency for Byzantine Total Order Broadcast via a new binary consensus called Blink, under partial synchrony with 5f+1 servers.
Introduces MFN selective memoization based on minimum feedback vertex set for linear-time backtracking regex matching and relates it to prior schemes under Thompson and Glushkov automata constructions.
An adaptive crawler using classifier feedback and seed-compound query generation discovers approximately 7.6 times more cumulative unique fake shopping hosts than a fixed-keyword baseline by cycle 3.
MIPSBLEED uncovers timing leaks in L1 data cache, L1 instruction cache, and execution engine of SMT-enabled MIPS processors and demonstrates a single-trace key recovery attack on elliptic curve cryptography.
citing papers explorer
-
On the Complexity of the Matching Problem of Regular Expressions with Backreferences
k-REWB matching cannot be solved in O(n to the 2k minus epsilon) time under SETH, is W[2]-hard parameterized by expression length, and 2-use 2-REWBs require superlinear time unless triangle detection does; 1-use REWBs admit an O(n log squared n) algorithm.
-
Defense against Poisoning Attacks under Shuffle-DP
A general framework defends any shuffle-DP protocol against poisoning attacks on union-preserving queries, retaining asymptotically equivalent error without attacks and only polylogarithmic increase with constant attackers.
-
SuperDP: Differential Privacy Refutation via Supermartingales
SuperDP refutes ε-DP via simultaneous synthesis of input pairs and witness functions using upper expectation supermartingales and lower expectation submartingales, delivering the first fully automated, sound, and semi-complete method applicable to both discrete and continuous stochastic mechanisms.
-
Open-World Assertion Checking for Smart Contracts via Game Semantics
A game semantics framework delivers the first sound and complete open-world assertion checking for Ethereum smart contracts, implemented as YulTracer which achieves perfect recall and precision on reentrancy benchmarks and real-world exploits.
-
CrypFormBench: Benchmarking Formal Analysis Capability of Large Language Models for Cryptographic Schemes
CrypFormBench is a new benchmark jointly covering symbolic and computational security to evaluate LLMs on five formal analysis capabilities, with results showing top model Claude-3.5 scores 48.7/100 and most models struggling on generation, transformation, and correction.
-
VIPIR: A Versatile GPU Framework for Integrating Private Information Retrieval Protocols
VIPIR introduces two new PIR protocols, ExpPack compression, and GPU optimizations for NTT and GEMM that deliver orders-of-magnitude higher throughput than prior systems.
-
Learning What to Forget: Improving LLM Unlearning via Learned Token-Level Importance
ATWU jointly optimizes model parameters and token weights via a linear scorer on hidden states, recovering oracle forget-specific tokens under a separation condition and achieving SOTA forget-retain trade-offs on TOFU and RWKU.
-
bpK#: Delegatable Pseudonyms And Their Applications to National eID Systems
Introduces bPk# as a delegatable pseudonym system with a formal framework, generic construction with security proofs, concrete instantiation, and reference implementation for eID applications.
-
A Bayesian Approach to Membership Inference for Statistical Release
A Bayesian decision framework with probabilistic programs computes membership posteriors for Bayesian network populations and outperforms marginal-based attacks on complex dependency structures.
-
Measuring the Depth of LLM Unlearning via Activation Patching
Introduces Unlearning Depth Score (UDS) via activation patching to quantify LLM unlearning depth and claims it outperforms 20 other metrics in faithfulness and robustness on 150 models.
-
Your Loss is My Gain: Low Stake Attacks on Liquid Staking Pools
A low-stake adversary can degrade a liquid staking pool's performance via consensus manipulation and profit from the resulting drop in its LST value through application-layer financial positions.
-
Compositional security definitions for higher-order where declassification
Constructs a logical-relations security model for where-declassification in higher-order languages by halting indistinguishability enforcement after relevant declassifications, yielding stronger guarantees than prior lower-order definitions.
-
Weaponizing the Commons: A Taxonomy and Detection Framework of Abuse on GitHub
A taxonomy of GitHub abuse behaviors is proposed along with a detection framework achieving F1-scores exceeding 89% on a manually labeled dataset of 392 instances.
-
Privatar: Scalable Privacy-preserving Multi-user VR via Secure Offloading
Privatar uses horizontal frequency partitioning and distribution-aware minimal perturbation to enable private offloading of VR avatar reconstruction, supporting 2.37x more users with modest overhead.
-
Typing Fallback Functions: A Semantic Approach to Type Safe Smart Contracts
Semantic typing via coinductively defined interpretations on a typed operational semantics ensures information flow control and non-interference for TinySol contracts that use fallback functions.
-
A Composable Game-Theoretic Framework for Blockchains
Introduces the first compositional game-theoretic framework for blockchains using cross-layer games and cross-application composition to analyze incentive compatibility and reveal vulnerabilities in composed systems.
-
Fast Byzantine Total Order Broadcast
Flutter achieves 2Δ + ε good-case latency for Byzantine Total Order Broadcast via a new binary consensus called Blink, under partial synchrony with 5f+1 servers.
-
Selective Memoization for Efficient Backtracking Regular Expression Matching
Introduces MFN selective memoization based on minimum feedback vertex set for linear-time backtracking regex matching and relates it to prior schemes under Thompson and Glushkov automata constructions.
-
Beyond Classification Accuracy: An Exploration-Range Evaluation of Adaptive Crawling for Fake Shopping Sites
An adaptive crawler using classifier feedback and seed-compound query generation discovers approximately 7.6 times more cumulative unique fake shopping hosts than a fixed-keyword baseline by cycle 3.
-
MIPSBLEED: Uncovering Microarchitectural Timing Leaks in Pervasive Embedded Processors
MIPSBLEED uncovers timing leaks in L1 data cache, L1 instruction cache, and execution engine of SMT-enabled MIPS processors and demonstrates a single-trace key recovery attack on elliptic curve cryptography.
-
A formal framework for the economic security of DeFi compositions
Introduces MEV non-interference and local MEV as formal notions to enable modular reasoning about economic security in DeFi contract compositions.
-
PeAR: A Static Binary Rewriting Framework for Binary-Only Fuzzing
PeAR shows static binary instrumentation can instrument 88% of FUZZBENCH targets with 4x throughput gains and coverage matching compiler-based methods.
-
EASE: Federated Multimodal Unlearning via Entanglement-Aware Anchor Closure
EASE closes three residual anchors in federated multimodal unlearning using bilateral displacement, cosine-sine decomposition, and forget lock, achieving near-retrain performance on forget and retain data.
-
CuLifter: Lifting GPU Binaries to Typed IR
CuLifter recovers types from untyped GPU register files via constraint propagation to lift 99.98% of 24,437 functions across 919 cubins to valid LLVM IR.
-
KindHML: formal verification of smart contracts based on Hennessy-Milner logic
An encoding of Solidity contracts and first-order Hennessy-Milner logic into Lustre enables Kind 2 model checking of complex temporal properties in smart contracts.
-
Sparse Concept Anchoring for Interpretable and Controllable Neural Representations
Sparse Concept Anchoring biases neural latent spaces toward targeted concepts using under 0.1% labels per concept, enabling reversible steering via projection and permanent removal via weight ablation with minimal side effects on other features.
-
Agents That Know Too Much: A Data-Centric Survey of Privacy in LLM Agents
A data-centric survey finds that only information-flow control covers compositional and cross-session leakage in LLM agents and that no single benchmark tests an agent across all its data surfaces under one policy.
-
CloakLM: Obfuscating GPU Memory Layout to Mitigate Model Ex-filtration for Serving
CloakLM mitigates model exfiltration by obfuscating GPU memory layouts with PCIe shaping, weight shuffling, and HBM remapping while keeping near-native performance.
-
To Wait or To Probe: Arbitrage Competition on High-Throughput Blockchains
On Base, probabilistic search is 23% of arbitrage activity but causes 95% of spam and 20% of gas use; protocol changes shift revenue toward successful trades and reduce spam share.
-
Rethinking Federated Unlearning via the Lens of Memorization
Introduces Grouped Memorization Evaluation and FedMemPrune to remove unique memorized information in federated unlearning while preserving overlapping knowledge.
-
Not Every Subject Should Stay: Machine Unlearning for Noisy Engagement Recognition
Approximate subject-level unlearning recovers 89.3% and 92.5% of oracle performance gains on EngageNet and DAiSEE at roughly one-quarter the retraining cost in K=3 forget-set regimes.
-
Adding Compilation Metadata To Binaries To Make Disassembly Decidable
Augmenting binaries with compilation metadata makes disassembly decidable, enables correct lifting to recompilable representations, and improves analysis reliability while adding negligible size and no runtime overhead.
-
How Generative AI Empowers Attackers and Defenders Across the Trust & Safety Landscape
Generative AI boosts attackers' ability to create harmful content at scale while also enabling defenders to detect threats, support users, and improve moderation processes.
-
Machine Unlearning for the XGBoost Model with Network Intrusion Datasets
XGBoost-Forget applies machine unlearning to XGBoost on IoT-23 and GeNIS network intrusion datasets, achieving faster forgetting with maintained predictive performance.
-
Projecting the Emerging Mindset of SWE Agent by Launching a Wild Code Understanding Journey
Ada is a scoped apparatus that records SWE-agent trajectories in real repositories and applies observation lenses to project navigation, evidence selection, synthesis, grounding, and stopping behaviors across 408 runs.
-
SoK: Post-Quantum Cryptography (PQC) Implementation in Software Systems
The paper synthesizes PQC implementation literature across human, organizational, and technological dimensions, identifies an imbalance favoring technology, and proposes the PQC-HOT conceptual model to guide coordinated socio-technical transitions.
-
Contextualizing Security and Privacy of Software-Defined Vehicles: A Literature Review and Industry Perspectives
A literature review and industry survey on SDV security and privacy produces a framework for addressing mixed-criticality systems, layered defenses, privacy techniques, and harmonized vehicle-cloud protections.
-
Authentication in Quantum Networks
A literature review of authentication in quantum networks concludes that it is not an intrinsic limitation but depends on explicit resources and deployment assumptions.