On-device research index

arXiv research

A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.

168,742 papers · 148 categories

Trend · papers per month

count % of corpus

4 results for “safety-relevant”

ALMANACS: A Simulatability Benchmark for Language Model Explainabilitycs.LG

ALMANACS benchmarks explainability methods on simulatability.

problem Evaluating the effectiveness of explainability methods for language models.

method ALMANACS is a simulatability benchmark that evaluates explainability methods on twelve safety-relevant topics.

result No explainability method outperforms the explanation-free control across all topics.

explainability language models benchmark simulatability safety-relevant

2023-12-20abs ↗pdf ↗

Uncertainty Wrapper in the medical domain: Establishing transparent uncertainty quantification for opaque machine learning models in practicecs.LG

Paper shows how to quantify uncertainty in medical ML models.

problem Uncertainty in opaque ML models can lead to safety risks in medical applications.

method Introduces Uncertainty Wrapper to quantify uncertainty transparently.

result Demonstrates practical utility of Uncertainty Wrapper in flow cytometry.

uncertainty quantification machine learning medical applications flow cytometry opaque models

2023-11-09abs ↗pdf ↗

Detection of False Positive and False Negative Samples in Semantic Segmentationcs.CV

The paper reviews techniques for detecting errors in semantic segmentation models.

problem Detecting false positives and false negatives in semantic segmentation models.

method Uncertainty quantification techniques applied to semantic segmentation.

result Techniques for detecting false positives and false negatives are proposed and discussed.

semantic segmentation false positives false negatives uncertainty quantification machine learning

2019-12-08abs ↗pdf ↗

Measuring Alignment-Induced Activation Shifts Correctly: A Template-Controlled Difference-in-Differences Protocolcs.LG

A new protocol corrects confounding effects to measure alignment-induced activation shifts accurately.

problem Confounding effects in measuring alignment-induced activation shifts using naive methods.

method Introduces a four-variant decomposition to separate alignment shift from template effects.

result Correctly measures alignment-induced activation shifts, recovering behaviorally active subspace.

activation shifts alignment difference-in-differences template control safety training

2026-05-23abs ↗pdf ↗