The emergence of foundation models in healthcare has opened new avenues for learning generalizable representations from large scale clinical data. Yet, existing approaches often struggle to reconcile the tabular and event based nature of Electronic Health Records (EHRs) with the sequential priors of natural language models. This structural mismatch limits their ability to capture longitudinal dependencies across patient encounters. We introduce SerialBEHRT, a domain aligned foundation model that extends SciBERT through additional pretraining on structured EHR sequences. SerialBEHRT is designed to encode temporal and contextual relationships among clinical events, thereby producing richer patient representations. We evaluate its effectiveness on the task of antibiotic susceptibility prediction, a clinically meaningful problem in antibiotic stewardship. Through extensive benchmarking against state of the art EHR representation strategies, we demonstrate that SerialBEHRT achieves superior and more consistent performance, highlighting the importance of temporal serialization in foundation model pretraining for healthcare.

Knowledge poisoning poses a critical threat to Retrieval-Augmented Generation (RAG) systems by injecting adversarial content into knowledge bases, tricking Large Language Models (LLMs) into producing attacker-controlled outputs grounded in manipulated context. Prior work highlights LLMs' susceptibility to misleading or malicious retrieved content. However, real-world fact-checking scenarios are more challenging, as credible evidence typically dominates the retrieval pool. To investigate this problem, we extend knowledge poisoning to the fact-checking setting, where retrieved context includes authentic supporting or refuting evidence. We propose \textbf{ADMIT} (\textbf{AD}versarial \textbf{M}ulti-\textbf{I}njection \textbf{T}echnique), a few-shot, semantically aligned poisoning attack that flips fact-checking decisions and induces deceptive justifications, all without access to the target LLMs, retrievers, or token-level control. Extensive experiments show that ADMIT transfers effectively across 4 retrievers, 11 LLMs, and 4 cross-domain benchmarks, achieving an average attack success rate (ASR) of 86\% at an extremely low poisoning rate of $0.93 \times 10^{-6}$, and remaining robust even in the presence of strong counter-evidence. Compared with prior state-of-the-art attacks, ADMIT improves ASR by 11.2\% across all settings, exposing significant vulnerabilities in real-world RAG-based fact-checking systems.

When does a large language model (LLM) know what it does not know? Uncertainty quantification (UQ) provides measures of uncertainty, such as an estimate of the confidence in an LLM's generated output, and is therefore increasingly recognized as a crucial component of trusted AI systems. Black-box UQ methods do not require access to internal model information from the generating LLM and therefore have numerous real-world advantages, such as robustness to system changes, adaptability to choice of LLM, reduced costs, and computational tractability. In this paper, we investigate the effectiveness of UQ techniques that are primarily but not necessarily entirely black-box, where the consistency between a generated output and other sampled generations is used as a proxy for confidence in its correctness. We propose a high-level non-verbalized similarity-based aggregation framework that subsumes a broad swath of UQ approaches suitable for complex generative tasks, as well as introduce specific novel techniques from the framework that train confidence estimation models using small training sets. Through an empirical study with datasets spanning the diverse tasks of question answering, summarization, and text-to-SQL, we demonstrate that our proposed similarity-based methods can yield better calibrated confidences than baselines.

Real-world data analysis tasks often come with under-specified goals and unclean data. User interaction is necessary to understand and disambiguate a user's intent, and hence, essential to solving these complex tasks. Existing benchmarks for evaluating LLMs on data analysis tasks do not capture these complexities or provide first-class support for interactivity. We introduce ConDABench, a framework for generating conversational data analysis (ConDA) benchmarks and evaluating external tools on the generated benchmarks. \bench consists of (a) a multi-agent workflow for generating realistic benchmarks from articles describing insights gained from public datasets, (b) 1,420 ConDA problems generated using this workflow, and (c) an evaluation harness that, for the first time, makes it possible to systematically evaluate conversational data analysis tools on the generated ConDA problems. Evaluation of state-of-the-art LLMs on the benchmarks reveals that while the new generation of models are better at solving more instances, they are not necessarily better at solving tasks that require sustained, long-form engagement. ConDABench is an avenue for model builders to measure progress towards truly collaborative models that can complete complex interactive tasks.

Transformer-based models have achieved remarkable performance in NLP tasks. However, their structural characteristics-multiple layers and attention heads-introduce efficiency challenges in inference and deployment. To address these challenges, various pruning methods have recently been proposed. Notably, gradient-based methods using Head Importance Scores (HIS) have gained traction for interpretability, efficiency, and ability to identify redundant heads. However, HIS alone has limitations as it captures only the gradient-driven contribution, overlooking the diversity of attention patterns. To overcome these limitations, we introduce a novel pruning criterion, HIES (Head Importance-Entropy Score), which integrates head importance scores with attention entropy, providing complementary evidence on per-head contribution. Empirically, HIES-based pruning yields up to 15.2% improvement in model quality and 2.04x improvement in stability over HIS-only methods, enabling substantial model compression without sacrificing either accuracy or stability. Code will be released upon publication.

The deployment of large language models (LLMs) in real-world applications is increasingly limited by their high inference cost. While recent advances in dynamic token-level computation allocation attempt to improve efficiency by selectively activating model components per token, existing methods rely on greedy routing--a myopic execute-or-skip mechanism that often leads to irreversible information loss and suboptimal token selection. This paper introduces informed routing, a new paradigm that proactively addresses these issues. The key insight is to assess not only a token's immediate importance but also its recoverability, i.e., how well its transformation can be approximated. To this end, we propose the Lightweight Feature Forecaster (LFF), a small predictive module that estimates a unit's output before routing decisions are made. This enables a flexible execute-or-approximate policy that preserves model fidelity while drastically reducing computation. Extensive experiments on both language modeling and reasoning tasks show that informed routing achieves state-of-the-art efficiency-performance trade-offs across multiple sparsity levels. Notably, even without final LoRA fine-tuning, our method matches or surpasses strong baselines that require full fine-tuning, all while reducing training time by over 50%. The emergence of large language models (LLMs) has catalyzed breakthroughs across diverse industries (Su et al., 2022; OpenAI et al., 2024; Rozi ` ere et al., 2024; Cai et al., 2025; Zheng et al., 2025).

Pairwise preference data have played an important role in the alignment of large language models (LLMs). Each sample of such data consists of a prompt, two different responses to the prompt, and a binary label indicating which of the two responses is better. The labels are usually annotated by professional human annotators. In this paper, we consider an alternative approach to collect pairwise preference data -- user annotation from comparison mode. With the increasingly wider adoption of LLMs among the population, users are contributing more and more of their preference labels through their daily interactions with the LLMs. The upside of such labels is that users are the best experts in judging the responses to their own queries/prompts, but the downside is the lack of quality control in these labels. In this paper, we consider a new idea of generating two responses from two different models or two different versions of the same model. The asymmetry allows us to make an inference of the user's data quality through our proposed user behavior model. We develop an expectation-maximization algorithm to estimate a latent quality factor of the user, and filter users' annotation data accordingly. The downstream task shows the effectiveness of our approach in both capturing the user behavior and data filtering for LLM alignment.

Explainable Artificial Intelligence (XAI) has been presented as the critical component for unlocking the potential of machine learning in mental health screening (MHS). However, a persistent lab-to-clinic gap remains. Current XAI techniques, such as SHAP and LIME, excel at producing technically faithful outputs such as feature importance scores, but fail to deliver clinically relevant, actionable insights that can be used by clinicians or understood by patients. This disconnect between technical transparency and human utility is the primary barrier to real-world adoption. This paper argues that this gap is a translation problem and proposes the Generative Operational Framework, a novel system architecture that leverages Large Language Models (LLMs) as a central translation engine. This framework is designed to ingest the raw, technical outputs from diverse XAI tools and synthesize them with clinical guidelines (via RAG) to automatically generate human-readable, evidence-backed clinical narratives. To justify our solution, we provide a systematic analysis of the components it integrates, tracing the evolution from intrinsic models to generative XAI. We demonstrate how this framework directly addresses key operational barriers, including workflow integration, bias mitigation, and stakeholder-specific communication. This paper also provides a strategic roadmap for moving the field beyond the generation of isolated data points toward the delivery of integrated, actionable, and trustworthy AI in clinical practice.

Artificial intelligence has advanced significantly through deep learning, reinforcement learning, and large language and vision models. However, these systems often remain task specific, struggle to adapt to changing conditions, and cannot generalize in ways similar to human cognition. Additionally, they mainly focus on mimicking brain structures, which often leads to black-box models with limited transparency and adaptability. Inspired by the structure and function of biological cognition, this paper introduces the concept of "Neurocognitive-Inspired Intelligence (NII)," a hybrid approach that combines neuroscience, cognitive science, computer vision, and AI to develop more general, adaptive, and robust intelligent systems capable of rapid learning, learning from less data, and leveraging prior experience. These systems aim to emulate the human brain's ability to flexibly learn, reason, remember, perceive, and act in real-world settings with minimal supervision. We review the limitations of current AI methods, define core principles of neurocognitive-inspired intelligence, and propose a modular, biologically inspired architecture that emphasizes integration, embodiment, and adaptability. We also discuss potential implementation strategies and outline various real-world applications, from robotics to education and healthcare. Importantly, this paper offers a hybrid roadmap for future research, laying the groundwork for building AI systems that more closely resemble human cognition.

The rapid expansion of IoT devices has outpaced current identification methods, creating significant risks for security, privacy, and network accountability. These challenges are heightened in open-world environments, where traffic metadata is often incomplete, noisy, or intentionally obfuscated. We introduce a semantic inference pipeline that reframes device identification as a language modeling task over heterogeneous network metadata. To construct reliable supervision, we generate high-fidelity vendor labels for the IoT Inspector dataset, the largest real-world IoT traffic corpus, using an ensemble of large language models guided by mutual-information and entropy-based stability scores. We then instruction-tune a quantized LLaMA3.18B model with curriculum learning to support generalization under sparsity and long-tail vendor distributions. Our model achieves 98.25% top-1 accuracy and 90.73% macro accuracy across 2,015 vendors while maintaining resilience to missing fields, protocol drift, and adversarial manipulation. Evaluation on an independent IoT testbed, coupled with explanation quality and adversarial stress tests, demonstrates that instruction-tuned LLMs provide a scalable and interpretable foundation for real-world device identification at scale.