Brief

Researchers have introduced MA-SBI, a novel framework for simulation-based inference that addresses challenges posed by simulator misspecification. Unlike previous methods requiring parameter calibration pairs, MA-SBI leverages unstructured side-channel information, such as text, to correct posterior estimates without retraining. The framework's theoretical bounds show that bias reduction is linked to the mutual information between misspecification and side-channel data. Empirical results demonstrate MA-SBI's effectiveness, matching oracle posteriors on benchmarks and improving performance on real-world epidemiological data. AI

Researchers are developing new methods to combat hallucinations in AI models, particularly in multimodal systems. One approach focuses on retrieval-augmented reliability-aware inference, which uses an external database to estimate prediction trustworthiness and abstain from answering when evidence is insufficient. Another method addresses semantic hallucination in explainable AI for vision-language models by disentangling unique semantic signals. Additionally, a technique called Attention Imbalance Rectification aims to reduce object hallucinations in Large Vision-Language Models by adjusting attention allocation. Finally, a study reformulates token-level hallucination detection as a quickest change detection problem to improve reaction time. AI