Search research

Agentic AI systems increasingly connect large language models (LLMs) to external scientific tools, yet whether and when tool access improves prediction accuracy remains uncharacterized. We present AGAPI (AtomGPT.org API), an open access platform integrating eight open-source LLMs with 18 REST endpoints (28 agent tools, 50 web apps) spanning materials databases, force fields, tight-binding band structures, X-ray diffraction, and protein structure. A three-evaluation residual decomposition on JARVIS-Leaderboard electronic-structure test sets separates agent pipeline fidelity from inherited density functional theory (DFT) functional bias. For bulk modulus and bandgap the agent reproduces JARVIS-DFT entries to numerical precision, so the experimental-reference degradation is functional bias, not agentic malfunction. On memorization-resistant test sets (57 defective supercells, 60 hypothetical compositions), tool-augmented mean absolute error (MAE) is below 0.005 eV versus 1.25 to 1.86 eV tool-free, confirming tools are indispensable where parametric knowledge is unavailable. We further demonstrate autonomous multi-step workflows including 10-operation defect-engineering pipelines. AGAPI is available at https://github.com/atomgptlab/agapi.

Generative AI offers new opportunities for individualized and adaptive learning, e.g., through large language model (LLM)-based feedback systems. While LLMs can produce factually correct feedback for relatively straightforward conceptual tasks, delivering high-quality feedback for tasks that require advanced domain expertise, such as physics problem solving, remains a substantial challenge. This study presents the design and implementation of an LLM-based feedback system for physics problem solving grounded in evidence-centered design and reports a first evaluation within the German Physics Olympiad. Participants rated the usefulness and correctness of the generated feedback for each implemented problem. The collected ratings indicate that the feedback was generally perceived as useful and highly correct. However, an in-depth analysis revealed that the feedback contained errors in 20% of cases; errors that often went unnoticed by the students. We discuss the risks associated with uncritical reliance on LLM-based feedback and outline potential directions for generating more adaptive and reliable LLM-based feedback in the future.

Recent advances in motion-aware large language models have shown remarkable promise for jointly learning motion understanding and generation knowledge. However, these models typically treat understanding and generation separately, limiting the mutual benefits that could arise from interactive feedback between tasks. In this work, we reveal that motion assessment and refinement tasks can act as crucial bridges to enable knowledge flow from motion understanding to generation. Specifically, we propose Interleaved Reasoning for Motion Generation (IRMoGen), a novel paradigm that tightly couples motion generation with assessment and refinement through iterative text-motion dialogue. To realize this, we introduce IRG-MotionLLM, the first model that seamlessly interleaves motion generation, assessment, and refinement to improve the alignment between generated motion and goal text. IRG-MotionLLM is developed progressively with a novel three-stage training scheme, initializing and subsequently enhancing native IRMoGen capabilities. To facilitate this development, we construct an automated data engine to synthesize interleaved reasoning annotations from existing text-motion datasets. Extensive experiments demonstrate the properties brought by IRMoGen training, and the advanced cross-benchmark and cross-evaluator performance of IRG-MotionLLM. Code and models are available at https://github.com/HumanMLLM/IRG-MotionLLM.

Large language model (LLM) agents exhibit strong mathematical problem-solving abilities and can even solve International Mathematical Olympiad (IMO) level problems with the assistance of formal proof systems. However, due to weak heuristics for auxiliary constructions, AI for geometry problem solving remains dominated by expert models such as AlphaGeometry 2, which rely heavily on large-scale data synthesis and search for both training and evaluation. In this work, we make the first attempt to build a medalist-level LLM agent for geometry and present InternGeometry. InternGeometry overcomes the heuristic limitations in geometry by iteratively proposing propositions and auxiliary constructions, verifying them with a symbolic engine, and reflecting on the engine's feedback to guide subsequent proposals. A dynamic memory mechanism enables InternGeometry to conduct more than two hundred interactions with the symbolic engine per problem. To further accelerate learning, we introduce Complexity-Boosting Reinforcement Learning (CBRL), which gradually increases the complexity of synthesized problems across training stages. Built on InternThinker-32B, InternGeometry solves 44 of 50 IMO geometry problems (2000-2024), exceeding the average gold medalist score (40.9), using only 13K training examples, just 0.004% of the data used by AlphaGeometry 2, demonstrating the potential of LLM agents on expert-level geometry tasks. InternGeometry can also propose novel auxiliary constructions for IMO problems that do not appear in human solutions.

Vision Language Models (VLMs) have shown promising planning capabilities, yet their success remains confined to the text domain, leaving visual decision-making relatively underexplored. Addressing this gap, we introduce Corrective Sequence Planning (CoSPlan) benchmark, where VLMs must plan a sequence of visual actions from an initial scene to a target scene. CoSPlan evaluates models on their ability to imagine and execute a coherent set of visual steps required to reach the goal (Step Completion). To prevent any shortcuts that simply describe the final scene, we introduce an erroneous action in decision-making, which must be detected (Error Detection) and corrected to reach the goal, enabling a deeper understanding of the task. CoSPlan spans across 4 tasks: maze navigation, block re-arrangement, image reconstruction, and object re-organization. Despite using advanced reasoning strategies such as Chain-of-Thought and Scene Graphs, VLMs struggle on CoSPlan, while still showing promising performance in the text domain. Addressing this, we propose Scene Graph Incremental updates (SGI), a novel training-free method to transform images into `textual' scene graphs, enabling step-by-step reasoning through iterative scene graph refinement. SGI yields an average of ~4.4% improvement on CoSPlan w/ generalization on PlanBench and VQA. Link for solving puzzles on the project page.

While Multimodal Large Language Models (MLLMs) have demonstrated significant promise in Vision-Language Navigation (VLN), existing agents remain heavily constrained by systemic bottlenecks across inference, training, and data collection. Specifically, they suffer from prohibitive latency due to visual history reprocessing, action leakage during sequence-packed training, and suboptimal exploration in self-correction data collection. To overcome these intertwined challenges, we present Efficient-VLN, a highly efficient and robust baseline that systematically resolves these issues through three simple-yet-effective mechanisms. (1) Inference: We introduce KV-cache reuse with contiguous RoPE, enabling the model to process only the newly observed frame at each step for real-time inference. (2) Training: We propose packed training with an action-isolating mask to accelerate throughput while effectively bridging the training-inference gap by preventing action leakage. (3) Data Collection: We employ an Adaptive DAgger to dynamically balance autonomous exploration and oracle guidance, enhancing error-recovery capability without escalating computational costs. Extensive evaluations show that Efficient-VLN significantly advances the state-of-the-art across the R2R-CE (73.2% SR) and RxR-CE (75.6% SR) benchmarks. Meanwhile, it yields a 28% latency reduction compared to the previous state-of-the-art StreamVLN, establishing a new paradigm for streaming MLLM-based navigation.

Reinforcement Learning from Human Feedback has emerged as a standard for aligning diffusion models. However, we identify a fundamental limitation in the standard DPO formulation because it relies on the Bradley-Terry model to aggregate diverse evaluation axes like aesthetic quality and semantic alignment into a single scalar reward. This aggregation creates a reward conflict where the model is forced to unlearn desirable features of a specific dimension if they appear in a globally non-preferred sample. To address this issue, we propose Multi Reward Conditional DPO (MCDPO). This method resolves reward conflicts by introducing a disentangled Bradley-Terry objective. MCDPO explicitly injects a preference outcome vector as a condition during training, which allows the model to learn the correct optimization direction for each reward axis independently within a single network. We further introduce dimensional reward dropout to ensure balanced optimization across dimensions. Extensive experiments on Stable Diffusion 1.5 and SDXL demonstrate that MCDPO achieves superior performance on benchmarks. Notably, our conditional framework enables dynamic and multiple-axis control at inference time using Classifier Free Guidance to amplify specific reward dimensions without additional training or external reward models.

LLMs have shown strong performance in data-rich domains such as programming, but their reliability in engineering tasks remains limited. Circuit analysis is particularly challenging because it requires both multimodal understanding and precise mathematical reasoning. This paper presents an enhanced end-to-end circuit problem-solving framework using Gemini 2.5 Pro as the backbone model for scalable engineering-education applications. We systematically evaluate Gemini 2.5 Pro on undergraduate circuit-analysis problems and identify two major failure modes: circuit-recognition hallucinations, especially source-polarity errors, and reasoning-process hallucinations, such as incorrect current-direction assumptions. To reduce recognition errors, we integrate a fine-tuned YOLO detector with OpenCV-based processing to isolate voltage and current sources for polarity re-identification. To mitigate reasoning errors, we introduce an ngspice-driven verification loop that supports iterative refinement with optional human feedback. On 83 problems, the proposed pipeline achieves 97.59% accuracy, compared with 79.52% for baseline Gemini. Across four hand-drawn diagram variations, accuracy improves from 60.61%--71.21% to 89.39%--92.42%, with statistically significant gains (p<0.005). On 43 problems from a different textbook, accuracy increases from 58.14% to 83.72%, further supporting cross-textbook generalizability. Error analysis shows that circuit recognition remains the dominant source of residual failures, particularly under varying diagram representations. Overall, the framework substantially improves the robustness, scalability, and generalizability of LLM-based circuit problem solving for engineering education and practical circuit analysis.

Monte Carlo Tree Search is a cornerstone algorithm for online planning, and its root-parallel variant is widely used when wall clock time is limited but best performance is desired. In environments with continuous action spaces, how to best aggregate statistics from different threads is an important yet underexplored question. In this work, we introduce a method that uses Gaussian Process Regression to obtain value estimates for promising actions that were not trialed in the environment. We perform a systematic evaluation across 6 different domains, demonstrating that our approach outperforms existing aggregation strategies while requiring a modest increase in inference time.

Quantum circuit design is a key bottleneck for practical quantum machine learning on complex, real-world data. We present an automated framework that discovers and refines variational quantum circuits (VQCs) using graph-based Bayesian optimization with a graph neural network (GNN) surrogate. Circuits are represented as graphs and mutated and selected via an expected improvement acquisition function informed by surrogate uncertainty with Monte Carlo dropout. Candidate circuits are evaluated with a hybrid quantum-classical variational classifier on the next generation firewall telemetry and network internet of things (NF-ToN-IoT-V2) cybersecurity dataset, after feature selection and scaling for quantum embedding. We benchmark our pipeline against an MLP-based surrogate, random search, and greedy GNN selection. The GNN-guided optimizer consistently finds circuits with lower complexity and competitive or superior classification accuracy compared to all baselines. Robustness is assessed via a noise study across standard quantum noise channels, including amplitude damping, phase damping, thermal relaxation, depolarizing, and readout bit flip noise. The implementation is fully reproducible, with time benchmarking and export of best found circuits, providing a scalable and interpretable route to automated quantum circuit discovery.

Neural decoding is fundamental to brain-computer interfaces, with growing applications in healthcare. Previous research has focused on leveraging signal processing and deep learning methods to enhance neural decoding performance. However, systematic guidance on architectural design for neural decoding remains limited. In this study, we develop NeuroSketch, a practical design recipe for neural decoding, through a basic architecture study followed by macro- and micro-level optimization. Comparing nine basic architectures, we find that CNN-2D outperforms other architectures in neural decoding tasks and explore its effectiveness from temporal and spatial perspectives. Building on this backbone, we combine gradual feature-map expansion and early downsampling at the macro level with grouped convolutions at the micro level. These choices form the recipe, which we instantiate as NeuroSketch-Base (1.4M parameters) and NeuroSketch-Large (4.2M parameters). The recipe is developed and evaluated through nearly 5,000 experiments across eight tasks spanning visual, auditory, and speech modalities and EEG, SEEG, and ECoG signals. Against ten baselines, the two variants collectively achieve the best accuracy on each task. Our code is available at https://github.com/Galaxy-Dawn/NeuroSketch.

Deep Operator Networks (DeepONets) have emerged as a powerful framework for data-driven operator learning, providing flexible surrogates for nonlinear mappings arising in partial differential equations (PDEs). However, the standard trunk network, which operates directly on raw spatial or spatiotemporal coordinates through fully connected layers, often struggles to represent sharp gradients, boundary layers, and other non-periodic solution structures on bounded domains. To address these limitations, we introduce the Spectral-Embedded Deep Operator Network (SEDONet), a novel DeepONet architecture in which the trunk is driven by a fixed Chebyshev spectral dictionary instead of coordinate inputs. This non-periodic spectral embedding provides a principled inductive bias for bounded domains, enabling the learned operator to capture fine-scale features that are difficult for Fourier-based or MLP-only trunks to represent. SEDONet is evaluated on the 2-D Poisson equation, 1-D Burgers' equation, 1-D advection-diffusion equation, Allen-Cahn equation, Lorenz-96 chaotic system, and Darcy flow, covering elliptic, hyperbolic, parabolic, chaotic, and multiscale problems. Across all benchmarks, SEDONet consistently achieves the lowest or statistically comparable relative $L^2$ errors among DeepONet, FEDONet, and SEDONet, with improvements of up to 54% over the baseline DeepONet and consistent gains over Fourier-embedded variants on bounded, non-periodic problems. Energy spectrum analyses further demonstrate that SEDONet more accurately preserves intermediate- and high-frequency solution structures. The proposed framework provides a simple, parameter-neutral modification to DeepONets, offering a robust and computationally efficient spectral approach for surrogate modeling of nonlinear operators in scientific computing.

Recent progress in text-to-video generation has achieved remarkable realism, yet fine-grained control over camera motion and orientation remains elusive, especially with extreme trajectories (e.g., a 180-degree turnaround, or looking directly up or down). Existing approaches typically encode camera trajectories using relative or ambiguous representations, limiting precise geometric control and offering limited support for large rotations. We introduce GimbalDiffusion, a framework that enables camera control grounded in physical-world coordinates, using gravity as a global reference. Instead of describing motion relative to previous frames, our method defines camera trajectories in an absolute coordinate system, allowing accurate, interpretable control over camera parameters. Using panoramic 360-degree videos for training, we cover the full sphere of possible viewpoints, including combinations of extreme pitch and roll that are out-of-distribution of conventional video data. To improve camera control, we introduce null-pitch conditioning, a strategy that prevents the model from overriding camera specifications in the presence of conflicting prompt content (e.g., generating grass while the camera points toward the sky). Finally, we propose new benchmarks to evaluate gravity-aware camera-controlled video generation, assessing models' ability to generate extreme camera angles and quantify their input prompt entanglement.

Vision-Language-Action (VLA) models have demonstrated remarkable capabilities in generalizing across diverse robotic manipulation tasks. However, deploying these models in unstructured environments remains challenging due to the critical need for simultaneous task compliance and safety assurance, particularly in preventing potential collisions during physical interactions. In this work, we introduce a Vision-Language-Safe Action (VLSA) architecture, named AEGIS, which contains a plug-and-play safety constraint (SC) layer formulated via control barrier functions. AEGIS integrates directly with existing VLA models to improve safety with theoretical guarantees, while maintaining their original instruction-following performance. To evaluate the efficacy of our architecture, we construct a comprehensive safety-critical benchmark SafeLIBERO, spanning distinct manipulation scenarios characterized by varying degrees of spatial complexity and obstacle intervention. Extensive experiments demonstrate the superiority of our method over state-of-the-art baselines. Notably, AEGIS achieves over 50% improvement in obstacle avoidance rate while substantially increasing the task success rate by nearly 10%. All benchmark datasets, code, and supplementary materials are publicly available at https://vlsa-aegis.github.io/.

Conditional diffusion models frequently suffer from language-image misalignments. Due to the ambiguity of intermediate noise corrupted latents, assessing prompt adherence currently requires completing the entire sampling trajectory. This late-stage evaluation incurs even higher computational costs during test-time scaling strategies, such as Best-of-N (BoN) sampling, as all misaligned trajectories must finish generation before being discarded. To tackle this, we propose NoisyCLIP, a noise-aware twin-tower model that enables early language-to-latent alignment estimation. By learning a vision encoder on noise-corrupted latents, we allow the model to "see" through the ambiguity of intermediate diffusion steps. To facilitate this training, we investigate noise-data augmentation sampling strategies and introduce two new benchmark datasets: Noisy-Conceptual-Captions and Noisy-GenAI-Bench. When applied as an early-stopping criterion for BoN, NoisyCLIP at half cost matches or beats frozen CLIP at full cost. Ultimately, this transforms alignment assessment from an expensive final check into a continuous monitoring tool, drastically reducing compute costs without sacrificing semantic fidelity.

Biosignals collected from wearable devices are widely utilized in healthcare applications. Machine learning models used in these applications often rely on features extracted from biosignals due to their effectiveness, lower data dimensionality, and wide compatibility across various model architectures. However, existing feature extraction methods often lack task-specific contextual knowledge, struggle to identify optimal features in high-dimensional combinatorial feature space, and are prone to automated code generation and execution errors. In this paper, we propose DeepFeature, the first LLM-empowered, context-aware feature generation framework for wearable biosignals. DeepFeature introduces a multi-source feature generation mechanism that integrates the inherent ability of LLMs, expert knowledge and inter-feature interactions. It also employs an iterative feature refinement process that uses feature assessment-based feedback for feature re-selection. Additionally, DeepFeature utilizes a robust multi-layer filtering and verification approach for feature description-to-code translation to ensure that the feature extraction functions run without crashing. Experimental evaluation results show that DeepFeature achieves the highest average AUROC across eight tasks under both sample-level and subject-level settings, outperforming the best baselines by 4.60% and 4.61%, respectively. DeepFeature achieves the most pronounced gains on the PPG-BP tasks, while remaining competitive with the best-performing baselines on Epilepsy, WESAD, and our self-collected SEN dataset.

Accurate and reliable aircraft landing time prediction is essential for effective resource allocation in air traffic management. However, the inherent uncertainty of aircraft trajectories and traffic flows poses significant challenges to both prediction accuracy and trustworthiness. Therefore, prediction models should not only provide point estimates of aircraft landing times but also the uncertainties associated with these predictions. Furthermore, aircraft trajectories are frequently influenced by the presence of nearby aircraft through air traffic control interventions such as radar vectoring. Consequently, landing time prediction models must account for multi-agent interactions in the airspace. In this work, we propose a probabilistic multi-agent aircraft landing time prediction framework that provides the landing times of multiple aircraft as distributions. We evaluate the proposed framework using an air traffic surveillance dataset collected from the terminal airspace of the Incheon International Airport in South Korea. The results demonstrate that the proposed model achieves higher prediction accuracy than the baselines and quantifies the associated uncertainties of its outcomes. In addition, the model uncovered underlying patterns in air traffic control through its attention scores, thereby enhancing explainability.

Camera-based animal re-identification (Animal Re-ID) can support wildlife monitoring and precision livestock management in large outdoor environments with limited wireless connectivity. In these settings, inference must run directly on collar tags or low-power edge nodes built around microcontrollers (MCUs), yet most Animal Re-ID models are designed for workstations or servers and are too large for devices with small memory and low-resolution inputs. We propose an on-device framework. First, we characterise the gap between state-of-the-art Animal Re-ID models and MCU-class hardware, showing that straightforward knowledge distillation from large teachers offers limited benefit once memory and input resolution are constrained. Second, guided by this analysis, we design a high-accuracy Animal Re-ID architecture by systematically scaling a CNN-based MobileNetV2 backbone for low-resolution inputs. Third, we evaluate the framework with a real-world dataset and introduce a data-efficient fine-tuning strategy to enable fast adaptation with just three images per animal identity at a new site. Across six public Animal Re-ID datasets, our compact model achieves competitive retrieval accuracy while reducing model size by over two orders of magnitude. On a self-collected cattle dataset, the deployed model performs fully on-device inference with only a small accuracy drop and unchanged Top-1 accuracy relative to its cluster version. We demonstrate that practical, adaptable Animal Re-ID is achievable on MCU-class devices, paving the way for scalable deployment in real field environments.

Reliable forecasting of multivariate time series under anomalous conditions is crucial in applications such as ATM cash logistics, where sudden demand shifts can disrupt operations. Modern deep forecasters achieve high accuracy on normal data but often fail when distribution shifts occur. We propose Weighted Contrastive Adaptation (WECA), a Weighted contrastive objective that aligns normal and anomaly-augmented representations, preserving anomaly-relevant information while maintaining consistency under benign variations. Evaluations on a nationwide ATM transaction dataset with domain-informed anomaly injection show that WECA improves SMAPE on anomaly-affected data by 6.1 percentage points compared to a normally trained baseline, with negligible degradation on normal data. These results demonstrate that WECA enhances forecasting reliability under anomalies without sacrificing performance during regular operations.

Radio frequency (RF)-based indoor localization offers significant promise for applications such as indoor navigation, augmented reality, and pervasive computing. While deep learning has greatly enhanced localization accuracy and robustness, existing localization models still face major challenges in cross-scene generalization due to their reliance on scene-specific labeled data. To address this, we introduce Radiance-Field Reinforced Pretraining (RFRP). This novel self-supervised pretraining framework couples a large localization model (LM) with a neural radio-frequency radiance field (RF-NeRF) in an asymmetrical autoencoder architecture. In this design, the LM encodes received RF spectra into latent, position-relevant representations, while the RF-NeRF decodes them to reconstruct the original spectra. This alignment between input and output enables effective representation learning using large-scale, unlabeled RF data, which can be collected continuously with minimal effort. To this end, we collected RF samples at 7,327,321 positions across 100 diverse scenes using four common wireless technologies--RFID, BLE, WiFi, and IIoT. Data from 75 scenes were used for training, and the remaining 25 for evaluation. Experimental results show that the RFRP-pretrained LM reduces localization error by over 40% compared to non-pretrained models and by 21% compared to those pretrained using supervised learning.

Deploying video anomaly detection (VAD) in the real world is often constrained by the scarcity, privacy, and cost of collecting real abnormal footage. We propose PA-VAD, a novel pseudo-only framework that trains an anomaly detector without using any real abnormal videos, by pairing real normal videos with diffusion-synthesized pseudo-abnormal videos generated from a small set of real normal images. Beyond proposing a generation-driven training pipeline, we make a key empirical discovery: pseudo anomalies exhibit a characteristic spatiotemporal magnitude bias in feature space, which can dominate Multiple Instance Learning and degrade generalization if left unaddressed. To counter this pseudo-induced bias, we introduce the Domain-Aligned Regularized Module (DARM), which combines domain alignment with usage-aware memory updates to balance prototype coverage and stabilize optimization under biased pseudo supervision. Extensive experiments demonstrate that PA-VAD achieves 98.2% AUC on ShanghaiTech, 82.5% on UCF-Crime, and 95.1% on XD-Violence, and further improves generalization to unseen anomaly classes in open-set evaluations. Notably, PA-VAD surpasses the best real-abnormal WVAD baselines on ShanghaiTech and XD-Violence by +0.6% and +0.9%, respectively, and improves over the UVAD state of the art on UCF-Crime by +1.9% -showing that high-accuracy VAD is attainable without collecting real abnormal videos.

Shadows encode rich information about scene geometry and illumination, yet existing methods either predict a unified shadow mask or overlook attached shadows entirely. We address this gap by proposing a framework for jointly detecting cast and attached shadows through explicit physical modeling of light direction and surface geometry under a dominant directional-light setting. Our approach is grounded in a simple observation: surfaces facing away from the light source tend to fall into shadow. We exploit the reciprocal relationship between shadow formation and light estimation to construct a closed feedback loop, a dual-module architecture in which a shadow detection module and a light estimation module iteratively refine each other. At each pass, updated light estimates, together with surface normals, produce partial attached shadow maps that guide detection, while improved shadow predictions sharpen light estimation. To support training and evaluation, we introduce a dataset of 1,458 images with manually annotated cast and attached shadow masks sourced from three existing benchmarks. Experiments demonstrate that our proposed method outperforms prior methods, with at least a 33% reduction in attached-shadow BER, while maintaining strong full-shadow and cast-shadow performance.

The visible parts of a scene are determined by occlusion among overlapping surfaces. Here we consider "dead leaves" models, which replicate this by independently sampling objects ("leaves") with position, shape, color, and texture and layering them until the image is covered. Building on prior theory, we present a self-contained framework that rigorously defines the dead leaves model and derives an analytical Bayesian ideal observer for partitioning finite pixel sets. The longest part of the paper spans the derivation of the prior probability, which elevates the observer beyond pixel-similarity methods by incorporating geometric information. These computations are practical only for small pixel sets (up to 9-10 pixels). We emphasize accessibility through step-by-step derivations, extensive visualizations, and examples. We empirically evaluate three tractable observers (prior-only, likelihood-only, and the full ideal observer), plus a random baseline on 108 dead leaves image datasets varying in texture intensity, leaf size, and image size. Likelihood-only performance falls with increasing texture intensity and image size. Prior-only performance falls with decreasing leaf size and increasing maximal image dimension. All model-based observers strongly outperform the random baseline, and the ideal observer consistently outperforms the others by combining both information sources. The model provides a principled upper bound on segmentation performance for limited pixel sets, enabling comparisons with human observers and algorithms.

Generative models have achieved remarkable progress with the emergence of flow matching (FM). It has demonstrated strong generative capabilities and attracted significant attention as a simulation-free flow-based framework capable of learning exact data densities. Motivated by these advances, we propose LatentFM, a flow-based model operating in the latent space for medical image segmentation. To model the data distribution, we first design two variational autoencoders (VAEs) to encode both medical images and their corresponding masks into a lower-dimensional latent space. We then estimate a conditional velocity field that guides the flow based on the input image. By sampling multiple latent representations, our method synthesizes diverse segmentation outputs whose pixel-wise variance reliably captures the underlying data distribution, enabling both highly accurate and uncertainty-aware predictions. Furthermore, we generate confidence maps that quantify the model certainty, providing clinicians with richer information for deeper analysis. We conduct experiments on two datasets, ISIC-2018 and CVC-Clinic, and compare our method with several prior baselines, including both deterministic and generative approach models. Through comprehensive evaluations, both qualitative and quantitative results show that our approach achieves superior segmentation accuracy while remaining highly efficient in the latent space.

Machine unlearning for large language models (LLMs) remains challenging because full retraining is costly, while approximate methods often struggle to remove targeted behaviors without degrading retained utility, especially under limited post-deployment supervision. We consider a practical PEFT setting for targeted behavioral contamination removal with a small forget set, a limited retain buffer, and LoRA-only updates, and propose RapidUn, an influence-guided framework that converts cross-sample influence estimates into fixed sample-specific weights for weighted LoRA unlearning. Across Llama-3-8B on Dolly-15k and Alpaca-57k, with cross-model validation on Mistral-7B + Dolly-15k, RapidUn achieves lower seen-trigger and OOD-trigger-family ASR than Fisher, GA, and LoReUn while maintaining competitive clean utility. On Llama-3-8B + Alpaca-57k, it achieves a 77x wall-clock speedup over the clean-corpus LoRA retraining reference. Complementary TOFU, semantic LLM-judge, and IFEval evaluations further support the effectiveness of influence-guided sample reweighting beyond the controlled trigger benchmark.

Joint video super-resolution and deblurring (VSRDB) requires both efficient long-range temporal modeling and robustness to frame-wise exposure-duration variation, which changes the extent of motion blur across video frames. We propose FMA-Net++, a non-recurrent, sequence-level framework built from Hierarchical Refinement with Bidirectional Aggregation (HRBA) blocks. By stacking HRBA blocks, FMA-Net++ processes video frames in parallel while hierarchically expanding the temporal receptive field, avoiding the limited temporal receptive field of sliding-window designs and the sequential bottleneck of recurrent ones. To handle exposure-duration-dependent blur, we introduce an Exposure Time-aware Modulation (ETM) layer that conditions HRBA features on exposure embeddings from an Exposure Time-aware Feature Extractor (ETE). The conditioned features guide an exposure-aware flow-guided dynamic filtering module to predict motion- and exposure-aware degradation kernels. FMA-Net++ decouples degradation learning from restoration: the former predicts degradation priors and the latter exploits them for efficient high-resolution restoration. To evaluate VSRDB under controlled exposure-duration variation, we introduce the REDS-ME (multi-exposure) and REDS-RE (random-exposure) benchmarks. Trained solely on synthetic data, FMA-Net++ achieves state-of-the-art accuracy and temporal consistency on these benchmarks. It further shows strong out-of-distribution performance on GoPro and challenging real-world videos, while outperforming recent methods in both restoration quality and inference speed.

Vision-language models (VLMs) such as CLIP are increasingly adapted across decentralized data silos, yet the reliability of their predictions under federated learning (FL) remains largely unexplored. In this work, we present a systematic study of calibration in federated CLIP under non-IID client distributions. Our experiments reveal that widely used prompt-tuning methods consistently degrade calibration, often yielding substantially higher calibration error despite competitive recognition performance, while explicit training-time calibration regularizers provide only limited improvements. Motivated by these findings, we identify the choice of fine-tuning parameterization as a critical factor governing calibration and conduct a controlled comparison between prompt tuning and five backbone fine-tuning (BFT) strategies: AdaptFormer, LayerNorm, LoRA, VeRA, and DoRA. We find that BFT methods generally offer a more favorable accuracy-calibration trade-off than prompt tuning, although their benefits are not universal. Through extensive analysis, we show that calibration behavior is closely linked to the geometry of federated updates, residual parameterization, and the resulting client and logit drift. Across in-distribution, domain-generalization, and base-to-new evaluation settings, our results establish fine-tuning parameterization as a central design choice for building accurate and reliable federated CLIP models. Codes are available at https://github.com/mainaksingha01/FL2oRA.

Targeted interventions on language models, such as unlearning or model editing, aim to modify specific information, but their effects often propagate to related, unintended areas (e.g., removing virology content may degrade performance on allergies); these side-effects are commonly referred to as the ripple effect. We introduce RippleBench-Maker, an automatic pipeline that retrieves semantic neighbors of any source concept from a knowledge repository and generates multiple-choice questions at varying semantic distances. We instantiate this framework using WikiRAG, an open-source RAG system over English Wikipedia, to construct RippleBench-WMDP-Bio (584 seed topics, 352,961 questions), and evaluate eight unlearning methods on Llama3-8B-Instruct. All eight exhibit accuracy drops that are largest near the unlearned target and decay with semantic distance, each with a distinct propagation profile. We replicate these findings across Mistral-7B, Zephyr-7B, and Yi-34B; cross-model delta curves are nearly identical, suggesting ripple effects are a property of the unlearning method rather than the base model. We validate all major pipeline stages using a four-experiment Mechanical Turk study (5,200+ responses, 61 workers). We release all code, data, and infrastructure.

As organizations increasingly deploy LLMs in sensitive domains such as legal, financial, and medical settings, ensuring alignment with internal organizational policies has become a priority. Existing content moderation frameworks remain largely confined to the safety domain and lack the robustness to capture nuanced organizational policies. LLM-as-a-judge and fine-tuning approaches, though flexible, introduce significant latency and training cost. To address these limitations, we frame policy violation detection as an out-of-distribution (OOD) problem in the model's activation space. We propose a training-free method that operates directly on the LLM internal representations, leveraging prior evidence that decision-relevant information is encoded within them. Inspired by whitening techniques, we derive policy-violation scores directly from normalized representations of LLM hidden activations. Our method requires only the policy text and a small number of illustrative samples, making it lightweight and easily deployable. We extensively evaluate our method across multiple LLMs and challenging policy benchmarks. It achieves up to 86.0% F1, outperforming both fine-tuned and LLM-as-a-judge baselines while requiring substantially less computation. Our code is publicly available at: https://github.com/FujitsuResearch/LLM-policy-violation-detection

Vibe coding is a new software development paradigm in which human engineers prompt a large language model (LLM) agent to complete complex coding tasks with little supervision. Although vibe coding is increasingly adopted, is the generated code really safe to deploy in production? To investigate this question, we propose SUSVIBES, a benchmark consisting of 186 feature-request software engineering tasks from real-world open-source projects, for which, human programmers committed vulnerable implementations. We evaluate 12 widely used coding agentic settings with frontier models on the benchmark. Disturbingly, all agents perform poorly in terms of software security. Although 57% of the solutions from SWE-Agent with Claude 4 Sonnet are functionally correct, only 11.8% are secure. Further experiments demonstrate that preliminary security strategies, such as augmenting the feature request with vulnerability hints, cannot mitigate these security issues. Our findings raise serious concerns about the widespread adoption of vibe coding, particularly in security-sensitive applications. The code and dataset are available at https://github.com/LeiLiLab/susvibes. The leaderboard is at https://leililab.github.io/susvibes-leaderboard.