Conformal Risk Control
PulseAugur coverage of Conformal Risk Control — every cluster mentioning Conformal Risk Control across labs, papers, and developer communities, ranked by signal.
3 day(s) with sentiment data
-
New OQRC Method Enhances ML Model Safety with Finite-Sample Guarantees
Researchers have introduced Occupancy-based Quantile Risk Control (OQRC), a new method designed to improve the safety and reliability of machine learning models. This approach extends existing conformal risk control fra…
-
New methods improve LLM judging in open-ended dialogue
Researchers have developed new methods for evaluating Large Language Models (LLMs) in open-ended dialogue scenarios. These methods, collectively termed Multi-Expert Conformal Risk Control (CRC), aim to improve the accur…
-
AI model for CT segmentation struggles with domain shift, researchers find
Researchers have developed a method for controlling risk in multi-organ CT segmentation, aiming to provide organ-specific recall guarantees for AI models. The study calibrated per-organ thresholds using an AMOS-trained …
-
New HG-CRC framework enhances LLM risk control across subgroups
Researchers have developed a new framework called Hierarchical Group-Conditional Conformal Risk Control (HG-CRC) to improve the reliability of large language models. This method ensures that risk guarantees are met not …
-
New framework assesses LLM output certifiability, identifies theoretical limits
Researchers have developed a framework to assess the certifiability of large language model (LLM) outputs for structured generation tasks like named-entity recognition and question answering. They established an impossi…
-
Federated Conformal Risk Control Protocol Addresses Hospital Data Vulnerabilities
Researchers have developed a new federated conformal risk control (CRC) protocol to address issues with standard CRC in multi-institutional deployments. The standard approach of pooling calibration scores across institu…
-
New method tackles foundation model risk under prompt and domain shifts
Researchers have developed PromptShift-CRC, a novel drift-aware conformal risk control method designed for foundation models facing evolving prompts and domain shifts. This method addresses the limitations of static cal…
-
New method optimizes power systems using decision-calibrated prediction sets
Researchers have developed a new method called decision-calibrated prediction sets for optimizing power system operations under uncertainty. This approach calibrates uncertainty sets based on the reliability of downstre…
-
Conf-Gen framework adapts conformal prediction for generative AI uncertainty
Researchers have introduced Conf-Gen, a new framework designed to adapt conformal risk control (CRC) for generative AI models. This method addresses the incompatibility of traditional conformal prediction (CP) with unsu…
-
AI model CRC-Screen certifies DNA synthesis orders against hazards
Researchers have developed a new method called CRC-Screen to improve the accuracy of screening DNA synthesis orders for potential hazards. The system addresses a critical flaw in current methods where screening fails wh…