S. Vosoughi

Unlocking Multilingual Reasoning Capabilities of LLMs and LVLMs via Representation Engineering

PRISM-BN: A Controlled Corpus and Benchmark for Text-to-Parameterized Bayesian Network Extraction

This work introduces PRISM-BN, a controlled corpus of 5054 BN-grounded descriptions paired with discrete reference BNs containing variables, states, directed edges, root priors, and full multi-parent CPDs across five domains and defines a benchmark with semantic node and state alignment, conditional structural scoring,...

Kernel-Complexity Edge Sanitization for Training-Free Defense against Structural Graph Attacks

Kernel-Complexity Edge Sanitization (KCES), a training-free and model-agnostic framework for defending against structural attacks, provides a principled and efficient framework for securing GNNs.

Improving Information Extraction with Learned Queries

This paper shows that another part of the pipeline matters at least as much: the queries used to elicit information extraction, and introduces List of Questions (LoQ), which generates document-specific question sets, and FeedQ, a feedback-driven optimization method that iteratively refines questions against extraction...

UTILMEM: Benchmarking Evidence Utilization in Long-Term Conversational Memory

UtilMem is introduced, a diagnostic benchmark comprising 1,717 instances across five domains, designed to evaluate four underexplored aspects of memory utilization: reasoning over dense histories, identifying implicitly relevant memories, synthesizing distributed evidence into summaries, analyses, or plans, and resisti...

When Linguistic and Internal Confidence Diverge in Large Language Models

Regression analyses show that distributional properties of confidence scores explain much of the observed alignment pattern, with model metadata playing a smaller role after controls, and support a lossy-channel view of linguistic confidence.