Large Language Models (LLMs) have been widely explored in educational scenarios. We identify a critical vulnerability in current educational LLMs, pedagogical jailbreaks, where students use answer-inducing prompts to elicit solutions rather than scaffolded instructions. To enable systematic study, we unify and formalize safe, helpful, and pedagogical behaviors with a knowledge-mastery graph and introduce SHAPE, a benchmark of 9,087 student-question pairs for evaluating tutoring behavior under adversarial pressure. We propose a graph-augmented tutoring pipeline that infers prerequisite concepts from queries, identifies mastery gaps, and routes generation between instructing and problem-solving via explicit gating. Experiments across multiple LLMs show that our method yields significantly improved safety under two pedagogical jailbreak settings, while maintaining near-ceiling helpfulness under the same evaluation protocol. Our code and data are available at https://github.com/MAPS-research/SHaPE

Large language models (LLMs) are increasingly utilized in clinical reasoning and risk assessment. However, their interpretive reliability in critical and indeterminate domains such as psychiatry remains unclear. Prior work has identified algorithmic biases and prompt sensitivity in these systems, raising concerns about how contextual information may influence model outputs, but there remains no systematic way to assess these, especially in the psychiatric domain. We propose an approach for reliability auditing downstream LLM tasks by structuring evaluation around the impact of prompt design and the inclusion of medically insignificant inputs on predicted hospitalization risk scores, which is often the first downstream AI clinical-decision-making task. In our audit, a cohort of synthetic patient profiles (n = 50) is generated, each consisting of 15 clinically relevant features and up to 50 clinically insignificant features, across four prompt reframings (neutral, logical, human impact, clinical judgment). We audit four LLMs (Gemini 2.5 Flash, LLaMa 3.3 70b, Claude Sonnet 4.6, GPT-4o mini), and our results show that including medically insignificant variables resulted in a statistically significant increase in the absolute mean predicted hospitalization risk and output variability across all models and prompts, indicating reduced predictive stability as contextual noise increased. Clinically insignificant features had an effect on instability across many model-prompt conditions, and prompt variations independently affected the trajectory of instability in a model-dependent manner. These findings quantify how LLM-based psychiatric risk assessments are sensitive to non-clinical information, highlighting the need for systematic evaluations of attributional stability and uncertainty behavior like this before clinical deployments.

Epidemic forecasting has become an integral part of real-time infectious disease outbreak response. While collaborative ensembles composed of statistical and machine learning models have become the norm for real-time forecasting, standardized benchmark datasets for evaluating such methods are lacking. Further, there is limited understanding on performance of these methods for novel outbreaks with limited historical data. In this paper, we propose IDOBE, a curated collection of epidemiological time series focused on outbreak forecasting. IDOBE compiles from multiple data repositories spanning over a century of surveillance and across U.S. states and global locations. We perform derivative-based segmentation to generate over 10,000 outbreaks covering multiple outcomes such as cases and hospitalizations for 13 diseases. We consider a variety of information-theoretic and distributional measures to quantify the epidemiological diversity of the dataset. Finally, we perform multi-horizon short-term forecasting (1- to 4-week-ahead) through the progression of the outbreak using 11 baseline models and report on their performance. In addition to standard metrics such as NMSE and MAPE for point forecasts, we include probabilistic scoring rules such as Normalized Weighted Interval Score (NWIS) to quantify the performance. We find that MLP-based methods have the most robust performance, with statistical methods having a slight edge during the pre-peak phase. IDOBE dataset along with baselines are released publicly on https://github.com/NSSAC/IDOBE to enable standardized, reproducible benchmarking of outbreak forecasting methods.

A central question in computational neuroscience is whether the learning rule used to train a neural network determines how well its internal representations align with those of the human visual cortex. We present a systematic comparison of four learning rules (backpropagation (BP), feedback alignment (FA), predictive coding (PC), and spike-timing-dependent plasticity (STDP)) applied to identical convolutional architectures and evaluated against human fMRI data from the THINGS-fMRI dataset (720 stimuli, 3 subjects) using Representational Similarity Analysis (RSA). All models process stimuli at 224 x 224 resolution; results are averaged across 5 random seeds. Crucially, we include an untrained random-weights baseline that reveals the dominant role of architecture. At V1/V2, the untrained baseline exceeds backpropagation (rho = 0.076 vs. rho = 0.034; Delta-rho = +0.044, p < 0.001), and STDP achieves the highest V1 alignment among trained rules (rho = 0.064). At LOC, only BP reliably exceeds the random baseline (rho = 0.012 vs. -0.005, p < 0.001). At IT, all five conditions converge (rho = 0.008-0.014) with no significant pairwise differences among trained rules (p > 0.05, FDR-corrected). FA consistently produces the lowest alignment at V1, V2, and LOC (rho = 0.012 at V1, below all other conditions). Partial RSA confirms all effects survive pixel-similarity control. Seed variability is small relative to between-rule differences at V1/V2. These results demonstrate that early visual alignment is architecture-driven, learning rules differentiate only at intermediate areas, and all rules converge at the highest levels of the hierarchy.

Dynamic 3D Gaussian Splatting methods achieve strong training-view PSNR on monocular video but generalize poorly: on the D-NeRF benchmark we measure an average train-test PSNR gap of 6.18 dB, rising to 11 dB on individual scenes. We report two findings that together account for most of that gap. Finding 1 (the role of splitting). A systematic ablation of the Adaptive Density Control pipeline (split, clone, prune, frequency, threshold, schedule) shows that splitting is responsible for over 80% of the gap: disabling split collapses the cloud from 44K to 3K Gaussians and the gap from 6.18 dB to 1.15 dB. Across all threshold-varying ablations, gap is log-linear in count (r = 0.995, bootstrap 95% CI [0.99, 1.00]), which suggests a capacity-based explanation. Finding 2 (the role of deformation coherence). We show that the capacity explanation is incomplete. A local-smoothness penalty on the per-Gaussian deformation field -- Elastic Energy Regularization (EER) -- reduces the gap by 40.8% while growing the cloud by 85%. Measuring per-Gaussian strain directly on trained checkpoints, EER reduces mean strain by 99.72% (median 99.80%) across all 8 scenes; on 8/8 scenes the median Gaussian under EER is less strained than the 1st-percentile (best-behaved) Gaussian under baseline. Alongside EER, we evaluate two further regularizers: GAD, a loss-rate-aware densification threshold, and PTDrop, a jitter-weighted Gaussian dropout. GAD+EER reduces the gap by 48%; adding PTDrop and a soft growth cap reaches 57%. We confirm that coherence generalizes to (a) a different deformation architecture (Deformable-3DGS, +40.6% gap reduction at re-tuned lambda), and (b) real monocular video (4 HyperNeRF scenes, reducing the mean PSNR gap by 14.9% at the same lambda as D-NeRF, with near-zero quality cost). The overfitting in dynamic 3DGS is driven by incoherent deformation, not parameter count.

Recent text-to-scene generation approaches largely reduced the manual efforts required to create 3D scenes. However, their focus is either to generate a scene layout or to generate objects, and few generate both. The generated scene layout is often simple even with LLM's help. Moreover, the generated scene is often inconsistent with the text input that contains non-trivial descriptions of the shape, appearance, and spatial arrangement of the objects. We present a new paradigm of sequential text-to-scene generation and propose a novel generative model for interactive scene creation. At the core is a 3D Autoregressive Diffusion model 3D-ARD+, which unifies the autoregressive generation over a multimodal token sequence and diffusion generation of next-object 3D latents. To generate the next object, the model uses one autoregressive step to generate the coarse-grained 3D latents in the scene space, conditioned on both the current seen text instructions and already synthesized 3D scene. It then uses a second step to generate the 3D latents in the smaller object space, which can be decoded into fine-grained object geometry and appearance. We curate a large dataset of 230K indoor scenes with paired text instructions for training. We evaluate 7B 3D-ARD+, on challenging scenes, and showcase the model can generate and place objects following non-trivial spatial layout and semantics prescribed by the text instructions.

Large Multimodal Models (LMMs) achieve state-of-the-art performance in high-stakes domains like healthcare, yet their reasoning remains opaque. Current interpretability methods, such as attention mechanisms or post-hoc saliency, often fail to faithfully represent the model's decision-making process, particularly when integrating heterogeneous modalities like time-series and text. We introduce Tree-of-Evidence (ToE), an inference-time search algorithm that frames interpretability as a discrete optimization problem. Rather than relying on soft attention weights, ToE employs lightweight Evidence Bottlenecks that score coarse groups or units of data (e.g., vital-sign windows, report sentences) and performs a beam search to identify the compact evidence set required to reproduce the model's prediction. We evaluate ToE across six tasks spanning three datasets and two domains: four clinical prediction tasks on MIMIC-IV, cross-center validation on eICU, and non-clinical fault detection on LEMMA-RCA. ToE produces auditable evidence traces while maintaining predictive performance, retaining over 0.98 of full-model AUROC with as few as five evidence units across all settings. Under sparse evidence budgets, ToE achieves higher decision agreement and lower probability fidelity error than other approaches. Qualitative analyses show that ToE adapts its search strategy: it often resolves straightforward cases using only vitals, while selectively incorporating text when physiological signals are ambiguous. ToE therefore provides a practical mechanism for auditing multimodal models by revealing which discrete evidence units support each prediction.

This paper presents a comprehensive review of the NTIRE 2026 3D Restoration and Reconstruction (3DRR) Challenge, detailing the proposed methods and results. The challenge seeks to identify robust reconstruction pipelines that are robust under real-world adverse conditions, specifically extreme low-light and smoke-degraded environments, as captured by our RealX3D benchmark. A total of 279 participants registered for the competition, of whom 33 teams submitted valid results. We thoroughly evaluate the submitted approaches against state-of-the-art baselines, revealing significant progress in 3D reconstruction under adverse conditions. Our analysis highlights shared design principles among top-performing methods and provides insights into effective strategies for handling 3D scene degradation.

Heterogeneity is a defining feature of deployed multi-robot teams: platforms often differ in sensing modalities, ranges, fields of view, and failure patterns. Controllers trained under nominal sensing can degrade sharply when deployed on robots with missing or mismatched sensors, even when the task and action interface are unchanged. We present DC-Ada, a reward-only decentralized adaptation method that keeps a pretrained shared policy frozen and instead adapts compact per-robot observation transforms to map heterogeneous sensing into a fixed inference interface. DC-Ada is gradient-free and communication-minimal: it uses budgeted accept/reject random search with short common-random-number rollouts under a strict step budget. We evaluate DC-Ada against four baselines in a deterministic 2D multi-robot simulator covering warehouse logistics, search and rescue, and collaborative mapping, across four heterogeneity regimes (H0--H3) and five seeds with a matched budget of $200{,}000$ joint environment steps per run. Results show that heterogeneity can substantially degrade a frozen shared policy and that no single mitigation dominates across all tasks and metrics. Observation normalization is strongest for reward robustness in warehouse logistics and competitive in search and rescue, while the frozen shared policy is strongest for reward in collaborative mapping. DC-Ada offers a useful complementary operating point: it improves completion most clearly in severe coverage-based mapping while requiring only scalar team returns and no policy fine-tuning or persistent communication. These results position DC-Ada as a practical deploy-time adaptation method for heterogeneous teams.

Assessing the veracity of a claim made online is a complex and important task with real-world implications. When these claims are directed at communities with limited access to information and the content concerns issues such as healthcare and culture, the consequences intensify, especially in low-resource languages. In this work, we introduce AfrIFact, a dataset that covers the necessary steps for automatic fact-checking (i.e., information retrieval, evidence extraction, and fact checking), in ten African languages and English. Our evaluation results show that even the best embedding models lack cross-lingual retrieval capabilities, and that cultural and news documents are easier to retrieve than healthcare-domain documents, both in large corpora and in single documents. We show that LLMs lack robust multilingual fact-verification capabilities in African languages, while few-shot prompting improves performance by up to 43% in AfriqueQwen-14B, and task-specific fine-tuning further improves fact-checking accuracy by up to 26%. These findings, along with our release of the AfrIFact dataset, encourage work on low-resource information retrieval, evidence retrieval, and fact checking.

Large Language Models (LLMs) achieve strong performance on many reasoning benchmarks, yet these evaluations typically focus on isolated tasks that differ from real-world usage in task-oriented dialogue (TOD). In this setting, LLMs must perform reasoning inherently while generating text and adhering to instructions on role, format, and style. This mismatch raises concerns about whether benchmark performance accurately reflects models' reasoning robustness in TOD setting. We investigate how framing reasoning tasks within TOD affects LLM performance by introducing BOULDER, a new dynamic benchmark covering eight travel-related tasks that require arithmetic, spatial, and temporal reasoning with both commonsense and formal aspects. Each problem is presented in both isolated and dialogue-based variants, enabling controlled comparison while mitigating data contamination. Experiments on eight LLMs reveal a substantial and consistent performance gap between isolated and dialogue settings. Through ablations and qualitative analysis, we show that this gap is largely driven by the multi-turn nature of dialogue, with additional effects from role conditioning and tool-use requirements. Our results highlight the need to evaluate LLM reasoning in realistic interactive scenarios.

Large language model (LLM) agents increasingly rely on external memory to support long-horizon interaction, personalized assistance, and multi-step reasoning. However, existing memory systems still face three core challenges: they often rely too heavily on semantic similarity, which can miss evidence crucial for user-centric understanding; they frequently store related experiences as isolated fragments, weakening temporal and causal coherence; and they typically use static memory granularities that do not adapt well to the requirements of different questions. We propose AdaMem, an adaptive user-centric memory framework for long-horizon dialogue agents. AdaMem organizes dialogue history into working, episodic, persona, and graph memories, enabling the system to preserve recent context, structured long-term experiences, stable user traits, and relation-aware connections within a unified framework. At inference time, AdaMem first resolves the target participant, then builds a question-conditioned retrieval route that combines semantic retrieval with relation-aware graph expansion only when needed, and finally produces the answer through a role-specialized pipeline for evidence synthesis and response generation. We evaluate AdaMem on the LoCoMo and PERSONAMEM benchmarks for long-horizon reasoning and user modeling. Experimental results show that AdaMem achieves state-of-the-art performance on both benchmarks. The code will be released upon acceptance.

Concept erasure aims to remove unwanted attributes, such as social or demographic factors, from learned representations, while preserving their task-relevant utility. While the goal of concept erasure is protection against all adversaries, existing methods remain vulnerable to nonlinear ones. This vulnerability arises from their failure to fully capture the complex, nonlinear statistical dependencies between learned representations and unwanted attributes. Moreover, although the existence of a trade-off between utility and erasure is expected, its progression during the erasure process, i.e., the cost of erasure, remains unstudied. In this work, we introduce Obliviator, a post-hoc erasure method designed to fully capture nonlinear statistical dependencies. We formulate erasure from a functional perspective, leading to an optimization problem involving a composition of kernels that lacks a closed-form solution. Instead of solving this problem in a single shot, we adopt an iterative approach that gradually morphs the feature space to achieve a more utility-preserving erasure. Unlike prior methods, Obliviator guards unwanted attribute against nonlinear adversaries. Our gradual approach quantifies the cost of nonlinear guardedness and reveals the dynamics between attribute protection and utility-preservation over the course of erasure. The utility-erasure trade-off curves obtained by Obliviator outperform the baselines and demonstrate its strong generalizability: its erasure becomes more utility-preserving when applied to the better-disentangled representations learned by more capable models.

Large language models are beginning to show steganographic capabilities. Such capabilities could allow misaligned models to evade oversight mechanisms. Yet principled methods to detect and quantify such behaviours are lacking. Classical definitions of steganography, and detection methods based on them, require a known reference distribution of non-steganographic signals. For the case of steganographic reasoning in LLMs, knowing such a reference distribution is not feasible; this renders these approaches inapplicable. We propose an alternative, \textbf{decision-theoretic view of steganography}. Our central insight is that steganography creates an asymmetry in usable information between agents who can and cannot decode the hidden content (present within a steganographic signal), and this otherwise latent asymmetry can be inferred from the agents' observable actions. To formalise this perspective, we introduce generalised $\mathcal{V}$-information: a utilitarian framework for measuring the amount of usable information within some input. We use this to define the \textbf{steganographic gap} -- a measure that quantifies steganography by comparing the downstream utility of the steganographic signal to agents that can and cannot decode the hidden content. We empirically validate our formalism, and show that it can be used to detect, quantify, and mitigate steganographic reasoning in LLMs.

Factorizable joint shift (FJS) represents a type of distribution shift (or dataset shift) that comprises both covariate and label shift. Recently, it has been observed that FJS actually arises from consecutive label and covariate (or vice versa) shifts. Research into FJS so far has been confined mostly to the case of categorical labels. We propose a framework for analysing distribution shift in the case of a general label space, thus covering both classification and regression models. Based on the framework, we generalise existing results on FJS to general label spaces and present and analyse a related extension to label distribution estimation of the expectation maximisation (EM) algorithm for class prior probabilities. We also take a fresh look at generalized label shift (GLS) in the case of a general label space.

The Third Perception Test challenge was organised as a full-day workshop alongside the IEEE/CVF International Conference on Computer Vision (ICCV) 2025. Its primary goal is to benchmark state-of-the-art video models and measure the progress in multimodal perception. This year, the workshop featured 2 guest tracks as well: KiVA (an image understanding challenge) and Physic-IQ (a video generation challenge). In this report, we summarise the results from the main Perception Test challenge, detailing both the existing tasks as well as novel additions to the benchmark. In this iteration, we placed an emphasis on task unification, as this poses a more challenging test for current SOTA multimodal models. The challenge included five consolidated tracks: unified video QA, unified object and point tracking, unified action and sound localisation, grounded video QA, and hour-long video QA, alongside an analysis and interpretability track that is still open for submissions. Notably, the unified video QA track introduced a novel subset that reformulates traditional perception tasks (such as point tracking and temporal action localisation) as multiple-choice video QA questions that video-language models can natively tackle. The unified object and point tracking merged the original object tracking and point tracking tasks, whereas the unified action and sound localisation merged the original temporal action localisation and temporal sound localisation tracks. Accordingly, we required competitors to use unified approaches rather than engineered pipelines with task-specific models. By proposing such a unified challenge, Perception Test 2025 highlights the significant difficulties existing models face when tackling diverse perception tasks through unified interfaces.

Current safety evaluations of large language models (LLMs) create a dangerous illusion of universal protection by aggregating harms under generic categories such as "Identity Hate", obscuring vulnerabilities toward specific populations. In this work, we expose the Selective Safety Trap: a systemic failure mode where models robustly defend specific populations while leaving underrepresented communities highly vulnerable to identical adversarial attacks. To systematically audit this phenomenon, we introduce MiJaBench, a bilingual (English-Portuguese) adversarial benchmark comprising 43,961 controlled jailbreaking prompts across 16 minority groups. By evaluating 14 state-of-the-art LLMs on MiJaBench, we curate 615,454 prompt-response pairs that compose MiJaBench-Align, revealing that safety alignment is not a uniform semantic capability but a demographic hierarchy, with defense rates fluctuating by up to 42% within the same model solely based on the target group. This disparity persists across architectures and languages and is amplified by scaling, indicating that current alignment methods learn group-specific safeguards rather than a generalized notion of harm. Through targeted direct preference optimization (DPO) on a 1B-parameter baseline, we achieve strong zero-shot safety generalizations to entirely unseen demographics and complex attack strategies. We release all datasets and scripts to provide the community with a concrete pathway toward equitable, transferable safety alignment.

Diffusion models have emerged as powerful priors for image editing tasks such as inpainting and local modification, where the objective is to generate realistic content that remains consistent with observed regions. In particular, zero-shot approaches that leverage a pretrained diffusion model, without any retraining, have been shown to achieve highly effective reconstructions. However, state-of-the-art zero-shot methods typically rely on a sequence of surrogate likelihood functions, whose scores are used as proxies for the ideal score. This procedure however requires vector-Jacobian products through the denoiser at every reverse step, introducing significant memory and runtime overhead. To address this issue, we propose a new likelihood surrogate that yields simple and efficient to sample Gaussian posterior transitions, sidestepping the backpropagation through the denoiser network. Our extensive experiments show that our method achieves strong observation consistency compared with fine-tuned baselines and produces coherent, high-quality reconstructions, all while significantly reducing inference cost. Code is available at https://github.com/YazidJanati/ding.

Conventional generative models for materials discovery are predominantly trained and validated using data from Density Functional Theory (DFT) with approximate exchange-correlation functionals. This creates a fundamental bottleneck: these models inherit DFT's systematic failures for strongly correlated systems, leading to exploration biases and an inability to discover materials where DFT predictions are qualitatively incorrect. We introduce a quantum-aware generative AI framework that systematically addresses this limitation through tight integration of multi-fidelity learning and active validation. Our approach employs a diffusion-based generator conditioned on quantum-mechanical descriptors and a validator using an equivariant neural network potential trained on a hierarchical dataset spanning multiple levels of theory (PBE, SCAN, HSE06, CCSD(T)). Crucially, we implement a robust active learning loop that quantifies and targets the divergence between low- and high-fidelity predictions. We conduct comprehensive ablation studies to deconstruct the contribution of each component, perform detailed failure mode analysis, and benchmark our framework against state-of-the-art generative models (CDVAE, GNoME, DiffCSP) across several challenging material classes. Our results demonstrate significant practical gains: a 3-5x improvement in successfully identifying potentially stable candidates in high-divergence regions (e.g., correlated oxides) compared to DFT-only baselines, while maintaining computational feasibility. This work provides a rigorous, transparent framework for extending the effective search space of computational materials discovery beyond the limitations of single-fidelity models.

We introduce StereoSpace, a diffusion-based framework for monocular-to-stereo synthesis that models geometry purely through viewpoint conditioning, without explicit depth or warping. A canonical rectified space and the conditioning guide the generator to infer correspondences and fill disocclusions end-to-end. To ensure fair and leakage-free evaluation, we introduce an end-to-end protocol that excludes any ground truth or proxy geometry estimates at test time. The protocol emphasizes metrics reflecting downstream relevance: iSQoE for perceptual comfort and MEt3R for geometric consistency. StereoSpace surpasses other methods from the warp & inpaint, latent-warping, and warped-conditioning categories, achieving sharp parallax and strong robustness on layered and non-Lambertian scenes. This establishes viewpoint-conditioned diffusion as a scalable, depth-free solution for stereo generation.

Vision-Language Models (VLMs) are essential for embodied AI and safety-critical applications, such as robotics and autonomous systems. However, existing benchmarks primarily focus on static or curated visual inputs, neglecting the challenges posed by adversarial conditions, value misalignment, and error propagation in continuous deployment. Current benchmarks either overlook the impact of real-world perturbations, or fail to account for the cumulative effect of inconsistent reasoning over time. To address these gaps, we introduce the Degraded Image Quality Leading to Hallucinations (DIQ-H) benchmark, the first to evaluate VLMs under adversarial visual conditions in continuous sequences. DIQ-H simulates real-world stressors including motion blur, sensor noise, and compression artifacts, and measures how these corruptions lead to persistent errors and misaligned outputs across time. The benchmark explicitly models error propagation and its long-term value consistency. To enhance scalability and reduce costs for safety-critical evaluation, we propose the Value-Guided Iterative Refinement (VIR) framework, which automates the generation of high-quality, ethically aligned ground truth annotations. VGIR leverages lightweight VLMs to detect and refine value misalignment, improving accuracy from 72.2% to 83.3%, representing a 15.3% relative improvement. The DIQ-H benchmark and VGIR framework provide a robust platform for embodied AI safety assessment, revealing vulnerabilities in error recovery, ethical consistency, and temporal value alignment.

Dynamic Bayesian networks (DBNs) are increasingly used in healthcare due to their ability to model complex temporal relationships in patient data while maintaining interpretability, an essential feature for clinical decision-making. However, existing approaches to handling missing data in longitudinal clinical datasets are largely derived from static Bayesian networks literature, failing to properly account for the temporal nature of the data. This gap limits the ability to quantify uncertainty over time, which is particularly critical in settings such as intensive care, where understanding the temporal dynamics is fundamental for model trustworthiness and applicability across diverse patient groups. Despite the potential of DBNs, a full Bayesian framework that integrates missing data handling remains underdeveloped. In this work, we propose a novel Gibbs sampling-based method for learning DBNs from incomplete data. Our method treats each missing value as an unknown parameter following a Gaussian distribution. At each iteration, the unobserved values are sampled from their full conditional distributions, allowing for principled imputation and uncertainty estimation. We evaluate our method on both simulated datasets and real-world intensive care data from critically ill patients. Compared to standard model-agnostic techniques such as MICE, our Bayesian approach demonstrates superior reconstruction accuracy and convergence properties. These results highlight the clinical relevance of incorporating full Bayesian inference in temporal models, providing more reliable imputations and offering deeper insight into model behavior. Our approach supports safer and more informed clinical decision-making, particularly in settings where missing data are frequent and potentially impactful.

Federated Learning (FL) enables collaborative model training without data sharing, yet participants face a fundamental challenge, e.g., simultaneously ensuring fairness across demographic groups while protecting sensitive client data. We introduce a differentially private fair FL algorithm (FedPF) that transforms this multi-objective optimization into a zero-sum game where fairness and privacy constraints compete against model utility. Our theoretical analysis reveals an inverse relationship: privacy mechanisms that protect sensitive attributes can reduce the statistical power available for detecting and correcting demographic biases under finite samples in federated settings. We further show that our theoretical bounds are consistent with a non-monotonic fairness-utility relationship, which is empirically validated by experiments where moderate fairness constraints improve generalization before excessive enforcement degrades performance. Compared with mainstream algorithms, even under strict privacy constraints, FedPF still maintains the lowest discrimination level among all tested algorithms while retaining high utility. Experimental validation demonstrates up to 42.9 % discrimination reduction across three datasets while maintaining competitive accuracy, but more importantly, reveals that achieving strong privacy and fairness simultaneously requires carefully balanced tradeoffs rather than optimizing either objective in isolation. Furthermore, hardware-level simulations demonstrate that FedPF maintains a low computational footprint, making it suitable for resource-constrained edge devices. The source code for our proposed algorithm is publicly accessible at https://github.com/szpsunkk/FedPF.

Machine Translation (MT) is widely employed to address resource scarcity in low-resource languages by generating synthetic data from high-resource counterparts. While sentiment preservation in translation has long been studied, a critical but underexplored factor is the role of cultural alignment between source and target languages. In this paper, we hypothesize that semantic labels are drifted or altered during MT due to cultural divergence. Through a series of experiments across culturally sensitive and neutral domains, we establish three key findings: (1) MT systems, including modern Large Language Models (LLMs), induce label drift during translation, particularly in culturally sensitive domains; (2) unlike earlier statistical MT tools, LLMs encode cultural knowledge, and leveraging this knowledge can amplify label drift; and (3) cultural similarity or dissimilarity between source and target languages is a crucial determinant of label preservation. Our findings highlight that neglecting cultural factors in MT not only undermines label fidelity but also risks misinterpretation and cultural conflict in downstream applications.

Embodied Reference Understanding requires identifying a target object in a visual scene based on both language instructions and pointing cues. While prior works have shown progress in open-vocabulary object detection, they often fail in ambiguous scenarios where multiple candidate objects exist in the scene. To address these challenges, we propose a novel ERU framework that jointly leverages LLM-based data augmentation, depth-map modality, and a depth-aware decision module. This design enables robust integration of linguistic and embodied cues, improving disambiguation in complex or cluttered environments. Experimental results on two datasets demonstrate that our approach significantly outperforms existing baselines, achieving more accurate and reliable referent detection.

Bayesian causal discovery benefits from prior information elicited from domain experts, and in heterogeneous domains any prior knowledge would be badly needed. However, so far prior elicitation approaches have assumed a single causal graph and hence are not suited to heterogeneous domains. We propose a causal elicitation strategy for heterogeneous settings, based on Bayesian experimental design (BED) principles, and a variational mixture structure learning (VaMSL) method -- extending the earlier differentiable Bayesian structure learning (DiBS) method -- to iteratively infer mixtures of causal Bayesian networks (CBNs). We construct an informative graph prior incorporating elicited expert feedback in the inference of mixtures of CBNs. Our proposed method successfully produces a set of alternative causal models (mixture components or clusters), and achieves an improved structure learning performance on heterogeneous synthetic data when informed by a simulated expert. Finally, we demonstrate that our approach is capable of capturing complex distributions in a breast cancer database.

We deploy large language models (LLMs) as business development (BD) agents for persuasive price negotiation in online travel agencies (OTAs). The agent must follow a multi-stage Standard Operating Procedure (SOP) and strict guardrails (no over-promising and no hallucinations), while remaining human-like and effective over long, multi-turn dialogues. We propose Reward-Enhanced Policy Optimization (REPO), a reinforcement learning post-training method that combines heterogeneous rewards: a preference-trained reward model (RM), an LLM-as-a-judge (RJ) for nuanced behaviors (e.g., emotional value and SOP compliance), and rule-based reward functions (RF) (mainly regex-based) for deterministic checks on numerics, formatting, and guardrails. In expert consensus evaluation (three human experts; 30 online conversations and 45 curated bad cases), REPO improves average dialogue rating to 4.63 (+0.33 over GRPO) and raises the share of conversations with at least one excellent response to 66.67% (+23.34 pp over GRPO), while achieving a 93.33% bad-case fix rate with 75.56% clean fixes. In a production A/B test on 9,653 real customer conversations (vs. an intent-driven dialogue system), REPO improves response rate by +12.14 pp and task success rate by +5.94 pp (p<0.001).

Developing effective prompts demands significant cognitive investment to generate reliable, high-quality responses from Large Language Models (LLMs). By deploying case-specific prompt engineering techniques that streamline frequently performed life sciences workflows, researchers could achieve substantial efficiency gains that far exceed the initial time investment required to master these techniques. The Prompt Report published in 2025 outlined 58 different text-based prompt engineering techniques, highlighting the numerous ways prompts could be constructed. To provide actionable guidelines and reduce the friction of navigating these various approaches, we distil this report to focus on 6 core techniques: zero-shot, few-shot approaches, thought generation, ensembling, self-criticism, and decomposition. We breakdown the significance of each approach and ground it in use cases relevant to life sciences, from literature summarization and data extraction to editorial tasks. We provide detailed recommendations for how prompts should and shouldn't be structured, addressing common pitfalls including multi-turn conversation degradation, hallucinations, and distinctions between reasoning and non-reasoning models. We examine context window limitations, agentic tools like Claude Code, while analyzing the effectiveness of Deep Research tools across OpenAI, Google, Anthropic and Perplexity platforms, discussing current limitations. We demonstrate how prompt engineering can augment rather than replace existing established individual practices around data processing and document editing. Our aim is to provide actionable guidance on core prompt engineering principles, and to facilitate the transition from opportunistic prompting to an effective, low-friction systematic practice that contributes to higher quality research.

Zero-Shot Video Anomaly Detection (ZS-VAD) requires temporally localizing anomalies without target domain training data, which is a crucial task due to various practical concerns, e.g., data privacy or new surveillance deployments. Skeleton-based approach has inherent generalizable advantages in achieving ZS-VAD as it eliminates domain disparities both in background and human appearance. However, existing methods only learn low-level skeleton representation and rely on the domain-limited normality boundary, which cannot generalize well to new scenes with different normal and abnormal behavior patterns. In this paper, we propose a novel zero-shot video anomaly detection framework, unlocking the potential of skeleton data via action typicality and uniqueness learning. Firstly, we introduce a language-guided semantic typicality modeling module that projects skeleton snippets into action semantic space and distills LLM's knowledge of typical normal and abnormal behaviors during training. Secondly, we propose a test-time context uniqueness analysis module to finely analyze the spatio-temporal differences between skeleton snippets and then derive scene-adaptive boundaries. Without using any training samples from the target domain, our method achieves state-of-the-art results against skeleton-based methods on four large-scale VAD datasets: ShanghaiTech, UBnormal, NWPU, and UCF-Crime, featuring over 100 unseen surveillance scenes.

Real-time streaming video understanding in domains such as autonomous driving and intelligent surveillance poses challenges beyond conventional offline video processing, requiring continuous perception, proactive decision making, and responsive interaction based on dynamically evolving visual content. However, existing methods rely on alternating perception-reaction or asynchronous triggers, lacking task-driven planning and future anticipation, which limits their real-time responsiveness and proactive decision making in evolving video streams. To this end, we propose a StreamAgent that anticipates the temporal intervals and spatial regions expected to contain future task-relevant information to enable proactive and goal-driven responses. Specifically, we integrate question semantics and historical observations through prompting the anticipatory agent to anticipate the temporal progression of key events, align current observations with the expected future evidence, and subsequently adjust the perception action (e.g., attending to task-relevant regions or continuously tracking in subsequent frames). To enable efficient inference, we design a streaming KV-cache memory mechanism that constructs a hierarchical memory structure for selective recall of relevant tokens, enabling efficient semantic retrieval while reducing the overhead of storing all tokens in the traditional KV-cache. Extensive experiments on streaming and long video understanding tasks demonstrate that our method outperforms existing methods in response accuracy and real-time efficiency, highlighting its practical value for real-world streaming scenarios.