The first dynamic algorithms for matrix rank and related objects achieve update times scaling with rank r, specifically Õ(r^1.405) per entry update and Õ(r^1.528 + z) per column update, extending to dynamic maximum matching.
Neural Cleanse: Identifying and mitigating back- door attacks in neural networks,
15 Pith papers cite this work, alongside 87 external citations. Polarity classification is still indexing.
15
Pith papers citing it
87
external citations · external index
citation-role summary
background 2
citation-polarity summary
roles
background 2polarities
background 2representative citing papers
A deterministic semi-streaming algorithm achieves an O(Δ)-coloring in O(√log Δ) passes, the first with linear palette size and sublogarithmic passes.
The authors derive the first bit-accurate arithmetic models for matrix multiply-accumulate operations on ten GPU architectures spanning NVIDIA Volta to Blackwell and AMD CDNA1 to CDNA3.
Chemistry-aware admission defeats many molecular graph backdoors, yet ChemBack demonstrates that chemically valid, target-aligned backdoors remain effective across benchmarks and defenses.
Argus enables backdoor detection in decentralized ML by collaborative neighbor-based validation of triggers, backed by convergence theory and reducing attack success by up to 90% on tested datasets.
First model organisms of narrow secret loyalties in LLMs evade black-box audits without principal knowledge and persist even at low poison fractions in training data.
Unsupervised GNN model learns local updates for approximate MaxIS on dynamic graphs, achieving competitive ratios on 200-1000 node instances and 1.00-1.18x larger solutions than other unsupervised models when generalizing to 100x larger graphs.
A gossip-native DAG consensus layer for decentralized federated learning certifies model-origin provenance via Hashgraph-style virtual voting, achieving learning quality comparable to blockchain-assisted FL with lower coordination overhead.
A composed pipeline of betweenness cuts, per-edge logistic classifiers, and GitHub-asserted edges resolves ~107M author identities while reducing the largest cluster from 170,431 to under 7,000 and improving gold recall from 0.44 to 0.70.
A scan of 5.8 billion git commits finds 17.59% carry cryptographic signatures, enabling a key-to-identity graph and trust tiers that calibrate heuristic author disambiguation.
Introduces ClawTrojan benchmark achieving 95.5% ASR for multi-step trojan attacks in agentic harnesses and DASGuard defense that sanitizes control content from untrusted sources.
C-mDP augments the secret with context, enforces metric indistinguishability on the augmented domain, and reduces the LP via conditional independence to achieve higher utility than standard mDP on vehicle traces.
GRASP reduces communication in remote control by 12-fold on average (50-fold for continuous actions) by having actors generate actions via guided sampling and local policy learning instead of receiving full actions or rewards.
A shared-commit Louvain deforking map for WoC V2604, with size caps and external GHArchive validation, recovers cross-forge fork families invisible to platform metadata graphs.
Module-switching defense disrupts backdoors more effectively than weight averaging with fewer models and remains robust even when some models share the same backdoors.
citing papers explorer
-
Dynamic Rank, Basis, and Matching
The first dynamic algorithms for matrix rank and related objects achieve update times scaling with rank r, specifically Õ(r^1.405) per entry update and Õ(r^1.528 + z) per column update, extending to dynamic maximum matching.
-
Faster Deterministic Streaming Vertex Coloring
A deterministic semi-streaming algorithm achieves an O(Δ)-coloring in O(√log Δ) passes, the first with linear palette size and sublogarithmic passes.
-
Bit-Accurate Modeling of GPU Matrix Multiply-Accumulate Units: Demystifying Numerical Discrepancy and Accuracy
The authors derive the first bit-accurate arithmetic models for matrix multiply-accumulate operations on ten GPU architectures spanning NVIDIA Volta to Blackwell and AMD CDNA1 to CDNA3.
-
Rethinking Molecular Graph Backdoors under Chemistry-aware Admission
Chemistry-aware admission defeats many molecular graph backdoors, yet ChemBack demonstrates that chemically valid, target-aligned backdoors remain effective across benchmarks and defenses.
-
Your Neighbors Know: Leveraging Local Neighborhoods for Backdoor Detection in Decentralized Learning
Argus enables backdoor detection in decentralized ML by collaborative neighbor-based validation of triggers, backed by convergence theory and reducing attack success by up to 90% on tested datasets.
-
Narrow Secret Loyalty Dodges Black-Box Audits
First model organisms of narrow secret loyalties in LLMs evade black-box audits without principal knowledge and persist even at low poison fractions in training data.
-
Unsupervised Learning of Local Updates for Maximum Independent Set in Dynamic Graphs
Unsupervised GNN model learns local updates for approximate MaxIS on dynamic graphs, achieving competitive ratios on 200-1000 node instances and 1.00-1.18x larger solutions than other unsupervised models when generalizing to 100x larger graphs.
-
Secure Decentralized Federated Learning via Gossip and Virtual Voting
A gossip-native DAG consensus layer for decentralized federated learning certifies model-origin provenance via Hashgraph-style virtual voting, achieving learning quality comparable to blockchain-assisted FL with lower coordination overhead.
-
Scaling Author Identity Disambiguation to the World of Code: A Methodology
A composed pipeline of betweenness cuts, per-edge logistic classifiers, and GitHub-asserted edges resolves ~107M author identities while reducing the largest cluster from 170,431 to under 7,000 and improving gold recall from 0.44 to 0.70.
-
Claimed or Attested? A Commit-Signature Dataset and Identity Trust Tiers across the World of Code
A scan of 5.8 billion git commits finds 17.59% carry cryptographic signatures, enabling a key-to-identity graph and trust tiers that calibrate heuristic author disambiguation.
-
From Prompt Injection to Persistent Control: Defending Agentic Harness Against Trojan Backdoors
Introduces ClawTrojan benchmark achieving 95.5% ASR for multi-step trojan attacks in agentic harnesses and DASGuard defense that sanitizes control content from untrusted sources.
-
Context-Aware Metric Differential Privacy for Vehicle Trajectory Data
C-mDP augments the secret with context, enforces metric indistinguishability on the augmented domain, and reduces the LP via conditional independence to achieve higher utility than standard mDP on vehicle traces.
-
Remote Action Generation: Remote Control with Minimal Communication
GRASP reduces communication in remote control by 12-fold on average (50-fold for continuous actions) by having actors generate actions via guided sampling and local policy learning instead of receiving full actions or rewards.
-
Deforking the World of Code: A Project-Provenance Map that Recovers Cross-Forge Fork Families that Platform Graphs Cannot See
A shared-commit Louvain deforking map for WoC V2604, with size caps and external GHArchive validation, recovers cross-forge fork families invisible to platform metadata graphs.
-
Defending against Backdoor Attacks via Module Switching
Module-switching defense disrupts backdoors more effectively than weight averaging with fewer models and remains robust even when some models share the same backdoors.