E-waste is growing rapidly while recycling rates remain low. We propose an electronic-device Graph-based Adaptive Planning (eGRAP) that integrates vision, dynamic planning, and dual-arm execution for autonomous disassembly. A camera-equipped arm identifies parts and estimates their poses, and a directed graph encodes which parts must be removed first. A scheduler uses topological ordering of this graph to select valid next steps and assign them to two robot arms, allowing independent tasks to run in parallel. One arm carries a screwdriver (with an eye-in-hand depth camera) and the other holds or handles components. We demonstrate eGRAP on 3.5in hard drives: as parts are unscrewed and removed, the system updates its graph and plan online. Experiments show consistent full disassembly of each HDD, with high success rates and efficient cycle times, illustrating the method's ability to adaptively coordinate dual-arm tasks in real time.

Data contamination undermines the validity of Large Language Model evaluation by enabling models to rely on memorized benchmark content rather than true generalization. While prior work has proposed contamination detection methods, these approaches are largely limited to English benchmarks, leaving multilingual contamination poorly understood. In this work, we investigate contamination dynamics in multilingual settings by fine-tuning several open-weight LLMs on varying proportions of Arabic datasets and evaluating them on original English benchmarks. To detect memorization, we extend the Tested Slot Guessing method with a choice-reordering strategy and incorporate Min-K% probability analysis, capturing both behavioral and distributional contamination signals. Our results show that translation into Arabic suppresses conventional contamination indicators, yet models still benefit from exposure to contaminated data, particularly those with stronger Arabic capabilities. This effect is consistently reflected in rising Mink% scores and increased cross-lingual answer consistency as contamination levels grow. To address this blind spot, we propose Translation-Aware Contamination Detection, which identifies contamination by comparing signals across multiple translated benchmark variants rather than English alone. The Translation-Aware Contamination Detection reliably exposes contamination even when English-only methods fail. Together, our findings highlight the need for multilingual, translation-aware evaluation pipelines to ensure fair, transparent, and reproducible assessment of LLMs.

Text-Based Person Search (TBPS) has seen significant progress with vision-language models (VLMs), yet it remains constrained by limited training data and the fact that VLMs are not inherently pre-trained for pedestrian-centric recognition. Existing TBPS methods therefore rely on dataset-centric fine-tuning to handle distribution shift, resulting in multiple independently trained models for different datasets. While synthetic data can increase the scale needed to fine-tune VLMs, it does not eliminate dataset-specific adaptation. This motivates a fundamental question: can we train a single unified TBPS model across multiple datasets? We show that naive joint training over all datasets remains sub-optimal because current training paradigms do not scale to a large number of unique person identities and are vulnerable to noisy image-text pairs. To address these challenges, we propose Scale-TBPS with two contributions: (i) a noise-aware unified dataset curation strategy that cohesively merges diverse TBPS datasets; and (ii) a scalable discriminative identity learning framework that remains effective under a large number of unique identities. Extensive experiments on CUHK-PEDES, ICFG-PEDES, RSTPReid, IIITD-20K, and UFine6926 demonstrate that a single Scale-TBPS model outperforms dataset-centric optimized models and naive joint training.

Modern machine learning systems are increasingly realised as multistage pipelines, yet existing transparency mechanisms typically operate at a model level: they describe what a system is and why it behaves as it does, but not how individual data samples are operationally recorded, tracked, and verified as they traverse the pipeline. This absence of verifiable, sample-level traceability leaves practitioners and users unable to determine whether a specific sample was used, when it was processed, or whether the corresponding records remain intact over time. We introduce FG-Trac, a model-agnostic framework that establishes verifiable, fine-grained sample-level traceability throughout machine learning pipelines. FG-Trac defines an explicit mechanism for capturing and verifying sample lifecycle events across preprocessing and training, computes contribution scores explicitly grounded in training checkpoints, and anchors these traces to tamper-evident cryptographic commitments. The framework integrates without modifying model architectures or training objectives, reconstructing complete and auditable data-usage histories with practical computational overhead. Experiments on a canonical convolutional neural network and a multimodal graph learning pipeline demonstrate that FG-Trac preserves predictive performance while enabling machine learning systems to furnish verifiable evidence of how individual samples were used and propagated during model execution.

We present VCNAC, a variable channel neural audio codec. Our approach features a single encoder and decoder parametrization that enables native inference for different channel setups, from mono speech to cinematic 5.1 channel surround audio. Channel compatibility objectives ensure that multi-channel content maintains perceptual quality when decoded to fewer channels. The shared representation enables training of generative language models on a single set of codebooks while supporting inference-time scalability across modalities and channel configurations. Evaluation using objective spatial audio metrics and subjective listening tests demonstrates that our unified approach maintains high reconstruction quality across mono, stereo, and surround audio configurations.

Unsupervised Environment Design (UED) seeks to automatically generate training curricula for reinforcement learning (RL) agents, with the goal of improving generalisation and zero-shot performance. However, designing effective curricula remains a difficult problem, particularly in settings where small subsets of environment parameterisations result in significant increases in the complexity of the required policy. Current methods struggle with a difficult credit assignment problem and rely on regret approximations that fail to identify challenging levels, both of which are compounded as the size of the environment grows. We propose Dynamic Environment Generation for UED (DEGen) to enable a denser level generator reward signal, reducing the difficulty of credit assignment and allowing for UED to scale to larger environment sizes. We also introduce a new regret approximation, Maximised Negative Advantage (MNA), as a significantly improved metric to optimise for, that better identifies more challenging levels. We show empirically that MNA outperforms current regret approximations and when combined with DEGen, consistently outperforms existing methods, especially as the size of the environment grows. We have made all our code available here: https://github.com/HarryMJMead/Dynamic-Environment-Generation-for-UED.

Time alters the visual appearance of entities in our world, like objects, places, and animals. Thus, for accurately generating contextually-relevant images, knowledge and reasoning about time can be crucial (e.g., for generating a landscape in spring vs. in winter). Yet, although substantial work exists on understanding and improving temporal knowledge in natural language processing, research on how temporal phenomena appear and are handled in text-to-image (T2I) models remains scarce. We address this gap with TempViz, the first data set to holistically evaluate temporal knowledge in image generation, consisting of 7.9k prompts and more than 600 reference images. Using TempViz, we study the capabilities of five T2I models across five temporal knowledge categories. Human evaluation shows that temporal competence is generally weak, with no model exceeding 75% accuracy across categories. Towards larger-scale studies, we also examine automated evaluation methods, comparing several established approaches against human judgments. However, none of these approaches provides a reliable assessment of temporal cues - further indicating the pressing need for future research on temporal knowledge in T2I.

Existing segmentation models exhibit significant vulnerability to adversarial attacks.To improve robustness, adversarial training incorporates adversarial examples into model training. However, existing attack methods consider only global semantic information and ignore contextual semantic relationships within the samples, limiting the effectiveness of adversarial training. To address this issue, we propose EroSeg-AT, a vulnerability-aware adversarial training framework that leverages EroSeg to generate adversarial examples. EroSeg first selects sensitive pixels based on pixel-level confidence and then progressively propagates perturbations to higher-confidence pixels, effectively disrupting the semantic consistency of the samples. Experimental results show that, compared to existing methods, our approach significantly improves attack effectiveness and enhances model robustness under adversarial training.

Large-scale Vision-Language-Action (VLA) models offer semantic generalization but suffer from high inference latency, limiting them to low-frequency batch-and-execute paradigm. This frequency mismatch creates an execution blind spot, causing failures in dynamic environments where targets move during the open-loop execution window. We propose TIDAL (Temporally Interleaved Diffusion and Action Loop), a hierarchical framework that decouples semantic reasoning from high-frequency actuation. TIDAL operates as a backbone-agnostic module for diffusion-based VLAs, using a dual-frequency architecture to redistribute the computational budget. Specifically, a low-frequency macro-intent loop caches semantic embeddings, while a high-frequency micro-control loop interleaves single-step flow integration with execution. This design enables approximately 9 Hz control updates on edge hardware (vs. approximately 2.4 Hz baselines) without increasing marginal overhead. To handle the resulting latency shift, we introduce a temporally misaligned training strategy where the policy learns predictive compensation using stale semantic intent alongside real-time proprioception. Additionally, we address the insensitivity of static vision encoders to velocity by incorporating a differential motion predictor. TIDAL is architectural, making it orthogonal to system-level optimizations. Experiments show a 2x performance gain over open-loop baselines in dynamic interception tasks. Despite a marginal regression in static success rates, our approach yields a 4x increase in feedback frequency and extends the effective horizon of semantic embeddings beyond the native action chunk size. Under non-paused inference protocols, TIDAL remains robust where standard baselines fail due to latency.

Semantic communication is emerging as a key enabler for distributed edge intelligence due to its capability to convey task-relevant meaning. However, achieving communication-efficient training and robust inference over wireless links remains challenging. This challenge is further exacerbated for multi-modal edge inference (MMEI) by two factors: 1) prohibitive communication overhead for distributed learning over bandwidth-limited wireless links, due to the \emph{multi-modal} nature of the system; and 2) limited robustness under varying channels and noisy multi-modal inputs. In this paper, we propose a three-stage communication-aware distributed learning framework to improve training and inference efficiency while maintaining robustness over wireless channels. In Stage~I, devices perform local multi-modal self-supervised learning to obtain shared and modality-specific encoders without device--server exchange, thereby reducing the communication cost. In Stage~II, distributed fine-tuning with centralized evidential fusion calibrates per-modality uncertainty and reliably aggregates features distorted by noise or channel fading. In Stage~III, an uncertainty-guided feedback mechanism selectively requests additional features for uncertain samples, optimizing the communication--accuracy tradeoff in the distributed setting. Experiments on RGB--depth indoor scene classification show that the proposed framework attains higher accuracy with far fewer training communication rounds and remains robust to modality degradation or channel variation, outperforming existing self-supervised and fully supervised baselines.

This study explores the capacity of generative artificial intelligence (Gen AI) to contribute to the construction of peace narratives and the revitalization of musical heritage in Mali. The study has been made in a political and social context where inter-community tensions and social fractures motivate a search for new symbolic frameworks for reconciliation. The study empirically explores three questions: (1) how Gen AI can be used as a tool for musical creation rooted in national languages and traditions; (2) to what extent Gen AI systems enable a balanced hybridization between technological innovation and cultural authenticity; and (3) how AI-assisted musical co-creation can strengthen social cohesion and cultural sovereignty. The experimental results suggest that Gen AI, embedded in a culturally conscious participatory framework, can act as a catalyst for symbolic diplomacy, amplifying local voices instead of standardizing them. However, challenges persist regarding the availability of linguistic corpora, algorithmic censorship, and the ethics of generating compositions derived from copyrighted sources.

Vision-Language Models (VLMs) enable multimodal reasoning for robotic perception and interaction, but their deployment in real-world systems remains constrained by latency, limited onboard resources, and privacy risks of cloud offloading. Edge intelligence within 6G, particularly Open RAN and Multi-access Edge Computing (MEC), offers a pathway to address these challenges by bringing computation closer to the data source. This work investigates the deployment of VLMs on ORAN/MEC infrastructure using the Unitree G1 humanoid robot as an embodied testbed. We design a WebRTC-based pipeline that streams multimodal data to an edge node and evaluate LLaMA-3.2-11B-Vision-Instruct deployed at the edge versus in the cloud under real-time conditions. Our results show that edge deployment preserves near-cloud accuracy while reducing end-to-end latency by 5\%. We further evaluate Qwen2-VL-2B-Instruct, a compact model optimized for resource-constrained environments, which achieves sub-second responsiveness, cutting latency by more than half but at the cost of accuracy.

Effective management of Type 1 Diabetes requires continuous glucose monitoring and precise insulin adjustments to prevent hyperglycemia and hypoglycemia. With the growing adoption of wearable glucose monitors and mobile health applications, accurate blood glucose prediction is essential for enhancing automated insulin delivery and decision-support systems. This paper presents a deep learning-based approach for personalized blood glucose prediction, leveraging patient-specific data to improve prediction accuracy and responsiveness in real-world scenarios. Unlike traditional generalized models, our method accounts for individual variability, enabling more effective subject-specific predictions. We compare Leave-One-Subject-Out Cross-Validation with a fine-tuning strategy to evaluate their ability to model patient-specific dynamics. Results show that personalized models significantly improve the prediction of adverse events, enabling more precise and timely interventions in real-world scenarios. To assess the impact of patient-specific data, we conduct experiments comparing a multimodal, patient-specific approach against traditional CGM-only methods. Additionally, we perform an ablation study to investigate model performance with progressively smaller training sets, identifying the minimum data required for effective personalization-an essential consideration for real-world applications where extensive data collection is often challenging. Our findings underscore the potential of adaptive, personalized glucose prediction models for advancing next-generation diabetes management, particularly in wearable and mobile health platforms, enhancing consumer-oriented diabetes care solutions.

Recent advances in large language models (LLMs) allow agents to represent actions as executable code, offering greater expressivity than traditional tool-calling. However, real-world tasks often demand both strategic planning and detailed implementation. Using a single agent for both leads to context pollution from debugging traces and intermediate failures, impairing long-horizon performance. We propose CodeDelegator, a multi-agent framework that separates planning from implementation via role specialization. A persistent Delegator maintains strategic oversight by decomposing tasks, writing specifications, and monitoring progress without executing code. For each sub-task, a new Coder agent is instantiated with a clean context containing only its specification, shielding it from prior failures. To coordinate between agents, we introduce Ephemeral-Persistent State Separation (EPSS), which isolates each Coder's execution state while preserving global coherence, preventing debugging traces from polluting the Delegator's context. Experiments on various benchmarks demonstrate the effectiveness of CodeDelegator across diverse scenarios.

Podcast script generation requires LLMs to synthesize structured, context-grounded dialogue from diverse inputs, yet systematic evaluation resources for this task remain limited. To bridge this gap, we introduce PodBench, a benchmark comprising 800 samples with inputs up to 21K tokens and complex multi-speaker instructions. We propose a multifaceted evaluation framework that integrates quantitative constraints with LLM-based quality assessment. Extensive experiments reveal that while proprietary models generally excel, open-source models equipped with explicit reasoning demonstrate superior robustness in handling long contexts and multi-speaker coordination compared to standard baselines. However, our analysis uncovers a persistent divergence where high instruction following does not guarantee high content substance. PodBench offers a reproducible testbed to address these challenges in long-form, audio-centric generation.

Recent debates on artificial intelligence increasingly emphasise questions of AI consciousness and moral status, yet there remains little agreement on how such properties should be evaluated. In this paper, we argue that awareness offers a more productive and methodologically tractable alternative. We introduce a practical method for evaluating awareness across diverse systems, where awareness is understood as encompassing a system's abilities to process, store and use information in the service of goal-directed action. Central to this approach is the claim that any evaluation aiming to capture the diversity of artificial systems must be domain-sensitive, deployable at any scale, multidimensional, and enable the prediction of task performance, while generalising to the level of abilities for the sake of comparison. Given these four desiderata, we outline a structured approach to evaluating and comparing awareness profiles across artificial systems with differing architectures, scales, and operational domains. By shifting the focus from artificial consciousness to being just aware enough, this approach aims to facilitate principled assessment, support design and oversight, and enable more constructive scientific and public discourse.

Background: Neuro-symbolic methods enhance the reliability of neural network classifiers through logical constraints, but they lack native support for ontologies. Objectives: We aim to develop a neuro-symbolic method that reliably outputs predictions consistent with a Description Logic ontology that formalizes domain-specific knowledge. Methods: We encode a Description Logic ontology as a circuit, a feed-forward differentiable computational graph that supports tractable execution of queries and transformations. We show that the circuit can be used to (i) generate synthetic datasets that capture the semantics of the ontology; (ii) efficiently perform deductive reasoning on a GPU; (iii) implement neuro-symbolic models whose predictions are approximately or provably consistent with the knowledge defined in the ontology. Results We show that the synthetic dataset generated using the circuit qualitatively captures the semantics of the ontology while being challenging for Machine Learning classifiers, including neural networks. Moreover, we show that compiling the ontology into a circuit is a promising approach for scalable deductive reasoning, with runtimes up to three orders of magnitude faster than available reasoners. Finally, we show that our neuro-symbolic classifiers reliably produce consistent predictions when compared to neural network baselines, maintaining competitive performances or even outperforming them. Conclusions By compiling Description Logic ontologies into circuits, we obtain a tighter integration between the Deep Learning and Knowledge Representation fields. We show that a single circuit representation can be used to tackle different challenging tasks closely related to real-world applications.

Reasoning models excel at complex tasks such as coding and mathematics, yet their inference is often slow and token-inefficient. To improve the inference efficiency, post-training quantization (PTQ) usually comes with the cost of large accuracy drops, especially for reasoning tasks under low-bit settings. In this study, we present a systematic empirical study of quantization-aware training (QAT) for reasoning models. Our key findings include: (1) Knowledge distillation is a robust objective for reasoning models trained via either supervised fine-tuning or reinforcement learning; (2) PTQ provides a strong initialization for QAT, improving accuracy while reducing training cost; (3) Reinforcement learning remains feasible for quantized models given a viable cold start and yields additional gains; and (4) Aligning the PTQ calibration domain with the QAT training domain accelerates convergence and often improves the final accuracy. Finally, we consolidate these findings into an optimized workflow (Reasoning-QAT), and show that it consistently outperforms state-of-the-art PTQ methods across multiple LLM backbones and reasoning datasets. For instance, on Qwen3-0.6B, it surpasses GPTQ by 44.53% on MATH-500 and consistently recovers performance in the 2-bit regime.

High-fidelity 4D dynamic facial avatar reconstruction from monocular video is a critical yet challenging task, driven by increasing demands for immersive virtual human applications. While Neural Radiance Fields (NeRF) have advanced scene representation, their capacity to capture high-frequency facial details, such as dynamic wrinkles and subtle textures from information-constrained monocular streams, requires significant enhancement. To tackle this challenge, we propose a novel hybrid neural radiance field framework, called Geometry-Aware-Transformer Enhanced NeRF (GAT-NeRF) for high-fidelity and controllable 4D facial avatar reconstruction, which integrates the Transformer mechanism into the NeRF pipeline. GAT-NeRF synergistically combines a coordinate-aligned Multilayer Perceptron (MLP) with a lightweight Transformer module, termed as Geometry-Aware-Transformer (GAT) due to its processing of multi-modal inputs containing explicit geometric priors. The GAT module is enabled by fusing multi-modal input features, including 3D spatial coordinates, 3D Morphable Model (3DMM) expression parameters, and learnable latent codes to effectively learn and enhance feature representations pertinent to fine-grained geometry. The Transformer's effective feature learning capabilities are leveraged to significantly augment the modeling of complex local facial patterns like dynamic wrinkles and acne scars. Comprehensive experiments unequivocally demonstrate GAT-NeRF's state-of-the-art performance in visual fidelity and high-frequency detail recovery, forging new pathways for creating realistic dynamic digital humans for multimedia applications.

Humanoid robots must adapt their contact behavior to diverse objects and tasks, yet most controllers rely on fixed, hand-tuned impedance gains and gripper settings. This paper introduces HumanoidVLM, a vision-language driven retrieval framework that enables the Unitree G1 humanoid to select task-appropriate Cartesian impedance parameters and gripper configurations directly from an egocentric RGB image. The system couples a vision-language model for semantic task inference with a FAISS-based Retrieval-Augmented Generation (RAG) module that retrieves experimentally validated stiffness-damping pairs and object-specific grasp angles from two custom databases, and executes them through a task-space impedance controller for compliant manipulation. We evaluate HumanoidVLM on 14 visual scenarios and achieve a retrieval accuracy of 93%. Real-world experiments show stable interaction dynamics, with z-axis tracking errors typically within 1-3.5 cm and virtual forces consistent with task-dependent impedance settings. These results demonstrate the feasibility of linking semantic perception with retrieval-based control as an interpretable path toward adaptive humanoid manipulation.

We introduce Strategic Doctrine Language Models (sdLM), a learning-system framework for multi-document strategic reasoning with doctrinal consistency constraints and calibrated uncertainty. The approach combines multi-document attention, temporal encoding, and a doctrine-consistency layer to improve long-horizon forecasting and plan plausibility while reducing severe doctrinal violations. We evaluate sdLM using (i) expert-panel scoring of strategic scenarios (N=47), (ii) doctrine consistency on 336 doctrine publications (12,847 statements), and (iii) geopolitical forecasting on 127 historical counterfactuals (1945-2020) across 12-60 month horizons. Across these benchmarks, sdLM achieves higher strategic quality and better calibration than strong general-purpose LLM baselines, and remains competitive with human experts on long-horizon judgments. We further report ablations, scaling trends, and deployment-oriented performance/latency characteristics to clarify which components drive improvements and how they translate to operational settings.

Memory-augmented language agents rely on embedding models for effective memory retrieval. However, existing training data construction overlooks a critical limitation: the hierarchical difficulty of negative samples and their natural distribution in human-agent interactions. In practice, some negatives are semantically close distractors while others are trivially irrelevant, and natural dialogue exhibits structured proportions of these types. Current approaches using synthetic or uniformly sampled negatives fail to reflect this diversity, limiting embedding models' ability to learn nuanced discrimination essential for robust memory retrieval. In this work, we propose a principled data construction framework HiNS that explicitly models negative sample difficulty tiers and incorporates empirically grounded negative ratios derived from conversational data, enabling the training of embedding models with substantially improved retrieval fidelity and generalization in memory-intensive tasks. Experiments show significant improvements: on LoCoMo, F1/BLEU-1 gains of 3.27%/3.30%(MemoryOS) and 1.95%/1.78% (Mem0); on PERSONAMEM, total score improvements of 1.19% (MemoryOS) and 2.55% (Mem0).

Microtubules are cytoskeletal filaments that play essential roles in many cellular processes and are key therapeutic targets in several diseases. Accurate segmentation of microtubule networks is critical for studying their organization and dynamics but remains challenging due to filament curvature, dense crossings, and image noise. We present MTFlow, a novel time-conditioned flow-matching model for microtubule segmentation. Unlike conventional U-Net variants that predict masks in a single pass, MTFlow learns vector fields that iteratively transport noisy masks toward the ground truth, enabling interpretable, trajectory-based refinement. Our architecture combines a U-Net backbone with temporal embeddings, allowing the model to capture the dynamics of uncertainty resolution along filament boundaries. We trained and evaluated MTFlow on synthetic and real microtubule datasets and assessed its generalization capability on public biomedical datasets of curvilinear structures such as retinal blood vessels and nerves. MTFlow achieves competitive segmentation accuracy comparable to state-of-the-art models, offering a powerful and time-efficient tool for filamentous structure analysis with more precise annotations than manual or semi-automatic approaches.

This paper introduces KRROOD, a framework designed to bridge the integration gap between modern software engineering and Knowledge Representation & Reasoning (KR&R) systems. While Object-Oriented Programming (OOP) is the standard for developing complex applications, existing KR&R frameworks often rely on external ontologies and specialized languages that are difficult to integrate with imperative code. KRROOD addresses this by treating knowledge as a first-class programming abstraction using native class structures, bridging the gap between the logic programming and OOP paradigms. We evaluate the system on the OWL2Bench benchmark and a human-robot task learning scenario. Experimental results show that KRROOD achieves strong performance while supporting the expressive reasoning required for real-world autonomous systems.

Soft robotic instruments could navigate delicate, tortuous anatomy more safely than rigid tools, but clinical adoption is limited by insufficient tip functionalization and real-time feedback at the tissue interface. Few sensing and therapeutic modules are compact, robust, and adaptable enough to measure, and respond to, subtle physiological cues during intraluminal procedures. We present a 1.47 mm diameter modular soft robotic catheter that integrates sensing, actuation, and therapy while retaining the compliance needed for safe endoluminal navigation. Validated across multiple in vivo settings, we emphasize its utility in endoscopic retrograde cholangiopancreatography (ERCP), a highly technical procedure and a key access route to the pancreas, an organ that is fragile, difficult to instrument, and central to diseases such as pancreatic cancer. Our architecture supports up to four independently controlled functional units, allowing customizable combinations of anchoring, manipulation, sensing, and targeted drug delivery. In a live porcine model, we demonstrate semi-autonomous deployment into the pancreatic duct and 7.5 cm of endoscopic navigation within it, a region currently inaccessible with standard catheters. A closed-loop autonomous/shared-control system that combines a learned model, magnetic actuation, onboard shape sensing, and visual marker tracking further improves cannulation accuracy. Together, these results establish a scalable platform for multifunctional soft robotic catheters and a new paradigm for complex endoluminal interventions, with potential to reduce radiation exposure, shorten training, and accelerate clinical translation of soft robotic technologies.

Vision-Language Models show strong zero-shot performance for chest X-ray classification, but standard flat metrics fail to distinguish between clinically minor and severe errors. This work investigates how to quantify and mitigate abstraction errors by leveraging medical taxonomies. We benchmark several state-of-the-art VLMs using hierarchical metrics and introduce Catastrophic Abstraction Errors to capture cross-branch mistakes. Our results reveal substantial misalignment of VLMs with clinical taxonomies despite high flat performance. To address this, we propose risk-constrained thresholding and taxonomy-aware fine-tuning with radial embeddings, which reduce severe abstraction errors to below 2 per cent while maintaining competitive performance. These findings highlight the importance of hierarchical evaluation and representation-level alignment for safer and more clinically meaningful deployment of VLMs.

As large language models (LLMs) are increasingly applied to creative writing, their performance on culturally specific narrative tasks warrants systematic investigation. This study constructs the first Chinese film script continuation benchmark comprising 53 classic films, and designs a multi-dimensional evaluation framework comparing GPT-5.2 and Qwen-Max-Latest. Using a "first half to second half" continuation paradigm with 3 samples per film, we obtained 303 valid samples (GPT-5.2: 157, 98.7% validity; Qwen-Max: 146, 91.8% validity). Evaluation integrates ROUGE-L, Structural Similarity, and LLM-as-Judge scoring (DeepSeek-Reasoner). Statistical analysis of 144 paired samples reveals: Qwen-Max achieves marginally higher ROUGE-L (0.2230 vs 0.2114, d=-0.43); however, GPT-5.2 significantly outperforms in structural preservation (0.93 vs 0.75, d=0.46), overall quality (44.79 vs 25.72, d=1.04), and composite scores (0.50 vs 0.39, d=0.84). The overall quality effect size reaches large effect level (d>0.8). GPT-5.2 excels in character consistency, tone-style matching, and format preservation, while Qwen-Max shows deficiencies in generation stability. This study provides a reproducible framework for LLM evaluation in Chinese creative writing.

In supervised learning with distributional inputs in the two-stage sampling setup, relevant to applications like learning-based medical screening or causal learning, the inputs (which are probability distributions) are not accessible in the learning phase, but only samples thereof. This problem is particularly amenable to kernel-based learning methods, where the distributions or samples are first embedded into a Hilbert space, often using kernel mean embeddings (KMEs), and then a standard kernel method like Support Vector Machines (SVMs) is applied, using a kernel defined on the embedding Hilbert space. In this work, we contribute to the theoretical analysis of this latter approach, with a particular focus on classification with distributional inputs using SVMs. We establish a new oracle inequality and derive consistency and learning rate results. Furthermore, for SVMs using the hinge loss and Gaussian kernels, we formulate a novel variant of an established noise assumption from the binary classification literature, under which we can establish learning rates. Finally, some of our technical tools like a new feature space for Gaussian kernels on Hilbert spaces are of independent interest.

Collaborative robots, or cobots, are increasingly integrated into various industrial and service settings to work efficiently and safely alongside humans. However, for effective human-robot collaboration, robots must reason based on human factors such as motivation level and aggression level. This paper proposes an approach for decision-making in human-robot collaborative (HRC) environments utilizing stochastic modeling. By leveraging probabilistic models and control strategies, the proposed method aims to anticipate human actions and emotions, enabling cobots to adapt their behavior accordingly. So far, most of the research has been done to detect the intentions of human co-workers. This paper discusses the theoretical framework, implementation strategies, simulation results, and potential applications of the bilateral collaboration approach for safety and efficiency in collaborative robotics.

Shape descriptors, i.e., per-vertex features of 3D meshes or point clouds, are fundamental to shape analysis. Historically, various handcrafted geometry-aware descriptors and feature refinement techniques have been proposed. Recently, several studies have initiated a new research direction by leveraging features from image foundation models to create semantics-aware descriptors, demonstrating advantages across tasks like shape matching, editing, and segmentation. Symmetry, another key concept in shape analysis, has also attracted increasing attention. Consequently, constructing symmetry-aware shape descriptors is a natural progression. Although the recent method $χ$ (Wang et al., 2025) successfully extracted symmetry-informative features from semantic-aware descriptors, its features are only one-dimensional, neglecting other valuable semantic information. Furthermore, the extracted symmetry-informative feature is usually noisy and yields small misclassified patches. To address these gaps, we propose a feature disentanglement approach which is simultaneously symmetry informative and symmetry agnostic. Further, we propose a feature refinement technique to improve the robustness of predicted symmetry informative features. Extensive experiments, including intrinsic symmetry detection, left/right classification, and shape matching, demonstrate the effectiveness of our proposed framework compared to various state-of-the-art methods, both qualitatively and quantitatively.