Introduces a 200-document benchmark and character-level R-Score for contextual PII redaction, with model evaluations and human agreement data showing the task remains unsolved.
lost-in-the-later
6 Pith papers cite this work. Polarity classification is still indexing.
6
Pith papers citing it
citation-role summary
background 2
citation-polarity summary
years
2026 6roles
background 2polarities
background 2representative citing papers
The paper demonstrates a black-box model extraction attack on graph classification models that leverages binary subgraph explanations to guide Monte Carlo edge sensitivity estimation with concentration guarantees.
NWCAD uses a two-stream setup with a two-stage gate to prevent accuracy drops on baseline-correct items under non-informative contexts while retaining gains from helpful contexts.
GRASP uses grounded chain-of-thought with a coordinate-weighted SFT stage and a GRPO-style policy-optimization stage to identify sarcasm targets in text and images, evaluated on the authors' new MSTI-MAX dataset.
DPGR from sequence counts maps to scaled differences in variant-specific Rt under a generation-interval model, with the equal-generation-time SIR case reducing to a direct scaled contrast.
A survey of reasoning language model adoption across 28 ERC scientific disciplines finds large maturity gaps, especially when only public resources are counted.
citing papers explorer
-
RedactionBench
Introduces a 200-document benchmark and character-level R-Score for contextual PII redaction, with model evaluations and human agreement data showing the task remains unsolved.
-
Can Subgraph Explanations Be Weaponized to Steal Graph Neural Networks?
The paper demonstrates a black-box model extraction attack on graph classification models that leverages binary subgraph explanations to guide Monte Carlo edge sensitivity estimation with concentration guarantees.
-
No-Worse Context-Aware Decoding: Preventing Neutral Regression in Context-Conditioned Generation
NWCAD uses a two-stream setup with a two-stage gate to prevent accuracy drops on baseline-correct items under non-informative contexts while retaining gains from helpful contexts.
-
GRASP: Grounded CoT Reasoning with Dual-Stage Optimization for Multimodal Sarcasm Target Identification
GRASP uses grounded chain-of-thought with a coordinate-weighted SFT stage and a GRPO-style policy-optimization stage to identify sarcasm targets in text and images, evaluated on the authors' new MSTI-MAX dataset.
-
On the Connection Between Differential Population Growth Rate and Epidemic Reproduction Numbers
DPGR from sequence counts maps to scaled differences in variant-specific Rt under a generation-interval model, with the equal-generation-time SIR case reducing to a direct scaled contrast.
-
Reasoning4Sciences: Bridging Reasoning Language Models to All Scientific Branches
A survey of reasoning language model adoption across 28 ERC scientific disciplines finds large maturity gaps, especially when only public resources are counted.