Search

Embodied manipulation is a fundamental ability in the realm of embodied artificial intelligence. Although current embodied manipulation models show certain generalizations in specific settings, they struggle in new environments and tasks due to the complexity and diversity of real-world scenarios. The traditional end-to-end data collection and training manner leads to significant data demands. Decomposing end-to-end tasks into atomic skills helps reduce data requirements and improves the task success rate. However, existing methods are limited by predefined skill sets that cannot be dynamically updated. To address the issue, we introduce a three-wheeled data-driven method to build an atomic skill library. We divide tasks into subtasks using the Vision-Language-Planning (VLP). Then, atomic skill definitions are formed by abstracting the subtasks. Finally, an atomic skill library is constructed via data collection and Vision-Language-Action (VLA) fine-tuning. As the atomic skill library expands dynamically with the three-wheel update strategy, the range of tasks it can cover grows naturally. In this way, our method shifts focus from end-to-end tasks to atomic skills, significantly reducing data costs while maintaining high performance and enabling efficient adaptation to new tasks. Extensive experiments in real-world settings demonstrate the effectiveness and efficiency of our approach.

Creating visually appealing advertising images is often a labor-intensive and time-consuming process. Is it possible to automatically generate such images using only basic product information--specifically, a product foreground image, taglines, and a target size? Existing methods mainly focus on parts of the problem and fail to provide a comprehensive solution. To address this gap, we propose a novel multistage framework called Product-Centric Advertising Image Design (PAID). It consists of four sequential stages to highlight product foregrounds and taglines while achieving overall image aesthetics: prompt generation, layout generation, background image generation, and graphics rendering. Different expert models are designed and trained for the first three stages: First, we use a visual language model (VLM) to generate background prompts that match the products. Next, a VLM-based layout generation model arranges the placement of product foregrounds, graphic elements (taglines and decorative underlays), and various nongraphic elements (objects from the background prompt). Following this, we train an SDXL-based image generation model that can simultaneously accept prompts, layouts, and foreground controls. To support the PAID framework, we create corresponding datasets with over 50,000 labeled images. Extensive experimental results and online A/B tests demonstrate that PAID can produce more visually appealing advertising images.

We present Step-Video-T2V, a state-of-the-art text-to-video pre-trained model with 30B parameters and the ability to generate videos up to 204 frames in length. A deep compression Variational Autoencoder, Video-VAE, is designed for video generation tasks, achieving 16x16 spatial and 8x temporal compression ratios, while maintaining exceptional video reconstruction quality. User prompts are encoded using two bilingual text encoders to handle both English and Chinese. A DiT with 3D full attention is trained using Flow Matching and is employed to denoise input noise into latent frames. A video-based DPO approach, Video-DPO, is applied to reduce artifacts and improve the visual quality of the generated videos. We also detail our training strategies and share key observations and insights. Step-Video-T2V's performance is evaluated on a novel video generation benchmark, Step-Video-T2V-Eval, demonstrating its state-of-the-art text-to-video quality when compared with both open-source and commercial engines. Additionally, we discuss the limitations of current diffusion-based model paradigm and outline future directions for video foundation models. We make both Step-Video-T2V and Step-Video-T2V-Eval available at https://github.com/stepfun-ai/Step-Video-T2V. The online version can be accessed from https://yuewen.cn/videos as well. Our goal is to accelerate the innovation of video foundation models and empower video content creators., Comment: 35 pages, 14 figures

Proteins are dynamic molecular machines whose biological functions, spanning enzymatic catalysis, signal transduction, and structural adaptation, are intrinsically linked to their motions. Designing proteins with targeted dynamic properties, however, remains a challenge due to the complex, degenerate relationships between sequence, structure, and molecular motion. Here, we introduce VibeGen, a generative AI framework that enables end-to-end de novo protein design conditioned on normal mode vibrations. VibeGen employs an agentic dual-model architecture, comprising a protein designer that generates sequence candidates based on specified vibrational modes and a protein predictor that evaluates their dynamic accuracy. This approach synergizes diversity, accuracy, and novelty during the design process. Via full-atom molecular simulations as direct validation, we demonstrate that the designed proteins accurately reproduce the prescribed normal mode amplitudes across the backbone while adopting various stable, functionally relevant structures. Notably, generated sequences are de novo, exhibiting no significant similarity to natural proteins, thereby expanding the accessible protein space beyond evolutionary constraints. Our work integrates protein dynamics into generative protein design, and establishes a direct, bidirectional link between sequence and vibrational behavior, unlocking new pathways for engineering biomolecules with tailored dynamical and functional properties. This framework holds broad implications for the rational design of flexible enzymes, dynamic scaffolds, and biomaterials, paving the way toward dynamics-informed AI-driven protein engineering.

Vision Transformers (ViTs) have outperformed traditional Convolutional Neural Network architectures and achieved state-of-the-art results in various computer vision tasks. Since ViTs are computationally expensive, the models either have to be pruned to run on resource-limited edge devices only or have to be executed on remote cloud servers after receiving the raw data transmitted over fluctuating networks. The resulting degraded performance or high latency all hinder their widespread applications. In this paper, we present Janus, the first framework for low-latency cloud-device collaborative Vision Transformer inference over dynamic networks. Janus overcomes the intrinsic model limitations of ViTs and realizes collaboratively executing ViT models on both cloud and edge devices, achieving low latency, high accuracy, and low communication overhead. Specifically, Janus judiciously combines token pruning techniques with a carefully designed fine-to-coarse model splitting policy and non-static mixed pruning policy. It attains a balance between accuracy and latency by dynamically selecting the optimal pruning level and split point. Experimental results across various tasks demonstrate that Janus enhances throughput by up to 5.15 times and reduces latency violation ratios by up to 98.7% when compared with baseline approaches under various network environments., Comment: Accepted for publication in IEEE INFOCOM 2025

We introduce Granite Vision, a lightweight large language model with vision capabilities, specifically designed to excel in enterprise use cases, particularly in visual document understanding. Our model is trained on a comprehensive instruction-following dataset, including document-related tasks, such as content extraction from tables, charts, diagrams, sketches, and infographics, as well as general image tasks. The architecture of Granite Vision is centered around visual modality alignment with a decoder-only, 2 billion parameter Granite large language model. Additionally, we introduce a dedicated safety classification approach in test-time that leverages a sparse set of attention vectors to identify potential harmful inputs. Despite its lightweight architecture, Granite Vision achieves strong results in standard benchmarks related to visual document understanding, as well as on the LiveXiv benchmark, which is designed to avoid test set contamination by using a constantly updated corpus of recently published Arxiv papers. We are releasing the model under the Apache-2 license, allowing for both research and commercial use, while offering complete visibility into the training data and other relevant details. See https://huggingface.co/ibm-granite/ for model weights.

Multi-agent reinforcement learning (MARL) lies at the heart of a plethora of applications involving the interaction of a group of agents in a shared unknown environment. A prominent framework for studying MARL is Markov games, with the goal of finding various notions of equilibria in a sample-efficient manner, such as the Nash equilibrium (NE) and the coarse correlated equilibrium (CCE). However, existing sample-efficient approaches either require tailored uncertainty estimation under function approximation, or careful coordination of the players. In this paper, we propose a novel model-based algorithm, called VMG, that incentivizes exploration via biasing the empirical estimate of the model parameters towards those with a higher collective best-response values of all the players when fixing the other players' policies, thus encouraging the policy to deviate from its current equilibrium for more exploration. VMG is oblivious to different forms of function approximation, and permits simultaneous and uncoupled policy updates of all players. Theoretically, we also establish that VMG achieves a near-optimal regret for finding both the NEs of two-player zero-sum Markov games and CCEs of multi-player general-sum Markov games under linear function approximation in an online environment, which nearly match their counterparts with sophisticated uncertainty quantification.

Answering questions with Chain-of-Thought (CoT) has significantly enhanced the reasoning capabilities of Large Language Models (LLMs), yet its impact on Large Multimodal Models (LMMs) still lacks a systematic assessment and in-depth investigation. In this paper, we introduce MME-CoT, a specialized benchmark evaluating the CoT reasoning performance of LMMs, spanning six domains: math, science, OCR, logic, space-time, and general scenes. As the first comprehensive study in this area, we propose a thorough evaluation suite incorporating three novel metrics that assess the reasoning quality, robustness, and efficiency at a fine-grained level. Leveraging curated high-quality data and a unique evaluation strategy, we conduct an in-depth analysis of state-of-the-art LMMs, uncovering several key insights: 1) Models with reflection mechanism demonstrate a superior CoT quality, with Kimi k1.5 outperforming GPT-4o and demonstrating the highest quality results; 2) CoT prompting often degrades LMM performance on perception-heavy tasks, suggesting a potentially harmful overthinking behavior; and 3) Although the CoT quality is high, LMMs with reflection exhibit significant inefficiency in both normal response and self-correction phases. We hope MME-CoT serves as a foundation for advancing multimodal reasoning in LMMs. Project Page: https://mmecot.github.io/, Comment: Project Page: https://mmecot.github.io/

We introduce a computationally efficient method for the automation of inverse design in science and engineering. Based on simple least-square regression, the underlying dynamic mode decomposition algorithm can be used to construct a low-rank subspace spanning multiple experiments in parameter space. The proposed inverse design dynamic mode composition (ID-DMD) algorithm leverages the computed low-dimensional subspace to enable fast digital design and optimization on laptop-level computing, including the potential to prescribe the dynamics themselves. Moreover, the method is robust to noise, physically interpretable, and can provide uncertainty quantification metrics. The architecture can also efficiently scale to large-scale design problems using randomized algorithms in the ID-DMD. The simplicity of the method and its implementation are highly attractive in practice, and the ID-DMD has been demonstrated to be an order of magnitude more accurate than competing methods while simultaneously being 3-5 orders faster on challenging engineering design problems ranging from structural vibrations to fluid dynamics. Due to its speed, robustness, interpretability, and ease-of-use, ID-DMD in comparison with other leading machine learning methods represents a significant advancement in data-driven methods for inverse design and optimization, promising a paradigm shift in how to approach inverse design in practice., Comment: 29 pages, 19 figures

Electric vehicle (EV) coordination can provide significant benefits through vehicle-to-everything (V2X) by interacting with the grid, buildings, and other EVs. This work aims to develop a V2X value-stacking framework, including vehicle-to-building (V2B), vehicle-to-grid (V2G), and energy trading, to maximize economic benefits for residential communities while maintaining distribution voltage. This work also seeks to quantify the impact of prediction errors related to building load, renewable energy, and EV arrivals. A dynamic rolling-horizon optimization (RHO) method is employed to leverage multiple revenue streams and maximize the potential of EV coordination. To address energy uncertainties, including hourly local building load, local photovoltaic (PV) generation, and EV arrivals, this work develops a Transformer-based forecasting model named Gated Recurrent Units-Encoder-Temporal Fusion Decoder (GRU-EN-TFD). The simulation results, using real data from Australia's National Electricity Market, and the Independent System Operators in New England and New York in the US, reveal that V2X value stacking can significantly reduce energy costs. The proposed GRU-EN-TFD model outperforms the benchmark forecast model. Uncertainties in EV arrivals have a more substantial impact on value-stacking performance, highlighting the significance of its accurate forecast. This work provides new insights into the dynamic interactions among residential communities, unlocking the full potential of EV batteries., Comment: 21 pages, accepted by Renewable Energy

One-shot Federated Learning (OFL) is a distributed machine learning paradigm that constrains client-server communication to a single round, addressing privacy and communication overhead issues associated with multiple rounds of data exchange in traditional Federated Learning (FL). OFL demonstrates the practical potential for integration with future approaches that require collaborative training models, such as large language models (LLMs). However, current OFL methods face two major challenges: data heterogeneity and model heterogeneity, which result in subpar performance compared to conventional FL methods. Worse still, despite numerous studies addressing these limitations, a comprehensive summary is still lacking. To address these gaps, this paper presents a systematic analysis of the challenges faced by OFL and thoroughly reviews the current methods. We also offer an innovative categorization method and analyze the trade-offs of various techniques. Additionally, we discuss the most promising future directions and the technologies that should be integrated into the OFL field. This work aims to provide guidance and insights for future research., Comment: 10 pages, 1 figure

We study the matter-wave solitons in Bose-Einstein condensate (BEC) trapped on a M\"{o}bius strip (MS), based on the respective Gross-Pitaevskii (GP) equation with the mean-field theory. In the linear regime, vortex states are characterized by quantum numbers, $n$ and $m$, corresponding to the transverse and circumferential directions, with the phase structure determined by the winding number (WN) $m$. Odd and even values of $n$ must associate, respectively, with integer and half-integer values of $m$, the latter ones requiring two cycles of motion around MS for returning to the initial phase. Using variational and numerical methods, we solve the GP equation with the attractive nonlinearity, producing a family of ground-state (GS) solitons for values of the norm below the critical one, above which the collapse sets in. Vortex solitons with $n=1,m=1$ and $% n=2,m=1/2$ are obtained in a numerical form. The vortex solitons with $% n=1,m=1$ are almost uniformly distributed in the azimuthal direction, while ones with $n=2,m=1/2$ form localized states. The Vakhitov-Kolokolov criterion and linear-stability analysis for the GS soliton solutions and vortices with $n=1,m=1$ demonstrates that they are completely stable, while the localized states with $n=2,m=1/2$ are completely unstable. Finally, the motion of solitons on the MS and the collision of two solitons are discussed., Comment: to be published in Physical Review A

The strong coupling between the spatial and polarisation degrees of freedom (DoF) in vector modes enables a diverse array of exotic, inhomogeneous polarisation distributions through a non-separable superposition, which are conventionally generated in circular-cylindrical symmetry. Here, we theoretically and experimentally demonstrate a generalized class of vector modes specified in elliptical spatial coordinates and elliptical polarisation. This generalisation gives rise to an even larger set of vector beams with more intricate polarisation distributions. Crucially, controlling the beam parameters allows engineering of vector beams with predefined polarisation trajectories on the Poincar\'e sphere. This capability offers potential applications, for example in optical communications, where precise polarisation control can significantly enhance data transmission and security.

Federated Learning (FL) allows users to collaboratively train a global machine learning model by sharing local model only, without exposing their private data to a central server. This distributed learning is particularly appealing in scenarios where data privacy is crucial, and it has garnered substantial attention from both industry and academia. However, studies have revealed privacy vulnerabilities in FL, where adversaries can potentially infer sensitive information from the shared model parameters. In this paper, we present an efficient masking-based secure aggregation scheme utilizing lightweight cryptographic primitives to mitigate privacy risks. Our scheme offers several advantages over existing methods. First, it requires only a single setup phase for the entire FL training session, significantly reducing communication overhead. Second, it minimizes user-side overhead by eliminating the need for user-to-user interactions, utilizing an intermediate server layer and a lightweight key negotiation method. Third, the scheme is highly resilient to user dropouts, and the users can join at any FL round. Fourth, it can detect and defend against malicious server activities, including recently discovered model inconsistency attacks. Finally, our scheme ensures security in both semi-honest and malicious settings. We provide security analysis to formally prove the robustness of our approach. Furthermore, we implemented an end-to-end prototype of our scheme. We conducted comprehensive experiments and comparisons, which show that it outperforms existing solutions in terms of communication and computation overhead, functionality, and security., Comment: 16 pages, 10 Figures

Due to their excellent drug-like and pharmacokinetic properties, small molecule drugs are widely used to treat various diseases, making them a critical component of drug discovery. In recent years, with the rapid development of deep learning (DL) techniques, DL-based small molecule drug discovery methods have achieved excellent performance in prediction accuracy, speed, and complex molecular relationship modeling compared to traditional machine learning approaches. These advancements enhance drug screening efficiency and optimization, and they provide more precise and effective solutions for various drug discovery tasks. Contributing to this field's development, this paper aims to systematically summarize and generalize the recent key tasks and representative techniques in DL-based small molecule drug discovery in recent years. Specifically, we provide an overview of the major tasks in small molecule drug discovery and their interrelationships. Next, we analyze the six core tasks, summarizing the related methods, commonly used datasets, and technological development trends. Finally, we discuss key challenges, such as interpretability and out-of-distribution generalization, and offer our insights into future research directions for DL-assisted small molecule drug discovery., Comment: 9 pages, 1 figures, 8 tables

A decisionmaker faces $n$ alternatives, each of which represents a potential reward. After investing costly resources into investigating the alternatives, the decisionmaker may select one, or more generally a feasible subset, and obtain the associated reward(s). The objective is to maximize the sum of rewards minus total costs invested. We consider this problem under a general model of an alternative as a "Markov Search Process," a type of undiscounted Markov Decision Process on a finite acyclic graph. Even simple cases generalize NP-hard problems such as Pandora's Box with nonobligatory inspection. Despite the apparently adaptive and interactive nature of the problem, we prove optimal prophet inequalities for this problem under a variety of combinatorial constraints. That is, we give approximation algorithms that interact with the alternatives sequentially, where each must be fully explored and either selected or else discarded before the next arrives. In particular, we obtain a computationally efficient $\frac{1}{2}-\epsilon$ prophet inequality for Combinatorial Markov Search subject to any matroid constraint. This result implies incentive-compatible mechanisms with constant Price of Anarchy for serving single-parameter agents when the agents strategically conduct independent, costly search processes to discover their values.

Unmanned Aerial Vehicles (UAVs) and Automated Guided Vehicles (AGVs) increasingly collaborate in logistics, surveillance, inspection tasks and etc. However, existing simulators often focus on a single domain, limiting cross-domain study. This paper presents the SkyRover, a modular simulator for UAV-AGV multi-agent pathfinding (MAPF). SkyRover supports realistic agent dynamics, configurable 3D environments, and convenient APIs for external solvers and learning methods. By unifying ground and aerial operations, it facilitates cross-domain algorithm design, testing, and benchmarking. Experiments highlight SkyRover's capacity for efficient pathfinding and high-fidelity simulations in UAV-AGV coordination. Project is available at https://sites.google.com/view/mapf3d/home., Comment: 9 pages

The recent Segment Anything Models (SAMs) have emerged as foundational visual models for general interactive segmentation. Despite demonstrating robust generalization abilities, they still suffer performance degradations in scenarios demanding accurate masks. Existing methods for high-precision interactive segmentation face a trade-off between the ability to perceive intricate local details and maintaining stable prompting capability, which hinders the applicability and effectiveness of foundational segmentation models. To this end, we present an SAM2Refiner framework built upon the SAM2 backbone. This architecture allows SAM2 to generate fine-grained segmentation masks for both images and videos while preserving its inherent strengths. Specifically, we design a localization augment module, which incorporates local contextual cues to enhance global features via a cross-attention mechanism, thereby exploiting potential detailed patterns and maintaining semantic information. Moreover, to strengthen the prompting ability toward the enhanced object embedding, we introduce a prompt retargeting module to renew the embedding with spatially aligned prompt features. In addition, to obtain accurate high resolution segmentation masks, a mask refinement module is devised by employing a multi-scale cascaded structure to fuse mask features with hierarchical representations from the encoder. Extensive experiments demonstrate the effectiveness of our approach, revealing that the proposed method can produce highly precise masks for both images and videos, surpassing state-of-the-art methods.

Recent advancements in autoregressive Large Language Models (LLMs) have achieved significant milestones, largely attributed to their scalability, often referred to as the "scaling law". Inspired by these achievements, there has been a growing interest in adapting LLMs for Recommendation Systems (RecSys) by reformulating RecSys tasks into generative problems. However, these End-to-End Generative Recommendation (E2E-GR) methods tend to prioritize idealized goals, often at the expense of the practical advantages offered by traditional Deep Learning based Recommendation Models (DLRMs) in terms of in features, architecture, and practices. This disparity between idealized goals and practical needs introduces several challenges and limitations, locking the scaling law in industrial RecSys. In this paper, we introduce a large user model (LUM) that addresses these limitations through a three-step paradigm, designed to meet the stringent requirements of industrial settings while unlocking the potential for scalable recommendations. Our extensive experimental evaluations demonstrate that LUM outperforms both state-of-the-art DLRMs and E2E-GR approaches. Notably, LUM exhibits excellent scalability, with performance improvements observed as the model scales up to 7 billion parameters. Additionally, we have successfully deployed LUM in an industrial application, where it achieved significant gains in an A/B test, further validating its effectiveness and practicality.

We summarize the status of lattice QCD ensemble generation efforts and their data management characteristics. Namely, these proceedings combine the contributions to a dedicated parallel session during the 41st International Symposium on Lattice Field Theory (Lattice 2024), during which representatives of 16 lattice QCD collaborations provided details on their simulation program, with focus on plans for publication, data management, and storage requirements. The parallel session was organized by the International Lattice Data Grid (ILDG), following an open call to the lattice QCD community for participation in the session., Comment: 20 pages. Proceedings summarizing the contributions to the "Lattice Data" session, held during the 41st International Symposium on Lattice Field theory (LATTICE2024), July 28th - August 3rd, 2024, Liverpool, UK

Sample selection is a prevalent approach in learning with noisy labels, aiming to identify confident samples for training. Although existing sample selection methods have achieved decent results by reducing the noise rate of the selected subset, they often overlook that not all mislabeled examples harm the model's performance equally. In this paper, we demonstrate that mislabeled examples correctly predicted by the model early in the training process are particularly harmful to model performance. We refer to these examples as Mislabeled Easy Examples (MEEs). To address this, we propose Early Cutting, which introduces a recalibration step that employs the model's later training state to re-select the confident subset identified early in training, thereby avoiding misleading confidence from early learning and effectively filtering out MEEs. Experiments on the CIFAR, WebVision, and full ImageNet-1k datasets demonstrate that our method effectively improves sample selection and model performance by reducing MEEs.

In this work, we develop a theoretical model for the cross-power spectrum of the galaxy density field before and after standard baryonic acoustic oscillation (BAO) reconstruction. Using this model, we extract the redshift-space distortion (RSD) parameter from the cross-power spectrum. The model is validated against a suite of high-resolution $N$-body simulations, demonstrating its accuracy and robustness for cosmological analyses., Comment: 14+6 pages, 6+1 figures

A transition metal diphosphide WP2 is a candidate for type-II Weyl semimetals (WSMs) in which spatial inversion symmetry is broken and Lorentz invariance is violated. As one of the key prerequisites for the presence of the WSM state in WP2, spatial inversion symmetry breaking in this compound has rarely been investigated by experiments. Furthermore, how much anisotropy the electrical properties of WP2 have and whether its electrical anisotropy can be tuned remain elusive. Here, we report angle-resolved polarized Raman spectroscopy, electrical transport, optical spectroscopy and first-principle studies of WP2. The energies of the observed Raman-active phonons and the angle dependences of the phonon intensities are well consistent with the results obtained by first-principle calculations and the analysis of the proposed crystal symmetry without spatial inversion, providing evidence that spatial inversion symmetry is broken in WP2. Moreover, the measured ratio (Rc/Ra) between the crystalline c-axis and a-axis electrical resistivities exhibits a weak dependence on temperature from 100 to 250 K, but increases abruptly below 100 K, and then reaches the value of 8.0 at 10 K, which is by far the strongest in-plane electrical resistivity anisotropy among the reported type-II WSM candidates with comparable carrier concentrations. Our optical-spectroscopy and calculation studies reveal that the abrupt enhancement of the Rc/Ra below 100 K mainly arises from a sharp increase in the scattering rate anisotropy at low temperatures. More interestingly, the Rc/Ra at 10 K can be tuned from 8.0 to 10.6 as the magnetic field increases from 0 to 9 T. The stronge and tunable electrical resistivity anisotropy found in WP2 can serve as a degree of freedom for tuning the electrical properties of type-II WSMs, which paves the way for developing novel electronic applications based on type-II WSMs., Comment: Published in Advanced Materials

Graph neural networks have been widely used in recent recommender systems, where negative sampling plays an important role. Existing negative sampling methods restrict the relationship between nodes as either hard positive pairs or hard negative pairs. This leads to the loss of structural information, and lacks the mechanism to generate positive pairs for nodes with few neighbors. To overcome limitations, we propose a novel soft link-based sampling method, namely MixDec Sampling, which consists of Mixup Sampling module and Decay Sampling module. The Mixup Sampling augments node features by synthesizing new nodes and soft links, which provides sufficient number of samples for nodes with few neighbors. The Decay Sampling strengthens the digestion of graph structure information by generating soft links for node embedding learning. To the best of our knowledge, we are the first to model sampling relationships between nodes by soft links in GNN-based recommender systems. Extensive experiments demonstrate that the proposed MixDec Sampling can significantly and consistently improve the recommendation performance of several representative GNN-based models on various recommendation benchmarks., Comment: 10 pages, 6 figures

This paper introduces Force Matching (ForM), a novel framework for generative modeling that represents an initial exploration into leveraging special relativistic mechanics to enhance the stability of the sampling process. By incorporating the Lorentz factor, ForM imposes a velocity constraint, ensuring that sample velocities remain bounded within a constant limit. This constraint serves as a fundamental mechanism for stabilizing the generative dynamics, leading to a more robust and controlled sampling process. We provide a rigorous theoretical analysis demonstrating that the velocity constraint is preserved throughout the sampling procedure within the ForM framework. To validate the effectiveness of our approach, we conduct extensive empirical evaluations. On the \textit{half-moons} dataset, ForM significantly outperforms baseline methods, achieving the lowest Euclidean distance loss of \textbf{0.714}, in contrast to vanilla first-order flow matching (5.853) and first- and second-order flow matching (5.793). Additionally, we perform an ablation study to further investigate the impact of our velocity constraint, reaffirming the superiority of ForM in stabilizing the generative process. The theoretical guarantees and empirical results underscore the potential of integrating special relativity principles into generative modeling. Our findings suggest that ForM provides a promising pathway toward achieving stable, efficient, and flexible generative processes. This work lays the foundation for future advancements in high-dimensional generative modeling, opening new avenues for the application of physical principles in machine learning.

By analyzing a $\psi(3686)$ data sample containing $(107.7\pm0.6)\times10^{6}$ events taken with the BESIII detector at the BEPCII storage ring in 2009, the $\chi_{c0}$ resonance parameters are precisely measured using $\chi_{c0,c2} \to \pi^+\pi^-/K^+K^-$ events. The mass of $\chi_{c0}$ is determined to be $M(\chi_{c0})=(3415.67\pm0.07\pm0.06\pm0.07$)~MeV/$c^2$, and its full width is $\Gamma(\chi_{c0})=(12.44\pm0.12\pm0.12)~{\rm MeV}$, where the first uncertainty is statistical, the second systematic, and the third for mass comes from $\chi_{c2}$ mass uncertainty. These measurements improve the precision of $\chi_{c0}$ mass by a factor of four and width by one order of magnitude over the previous individual measurements, and significantly boost our knowledge about the charmonium spectrum. Together with additional $(345.4\pm2.6)\times10^{6}$ $\psi(3686)$ data events taken in 2012, the decay branching fractions of $\chi_{c0,c2}\to\pi^+\pi^-/K^+K^-$ are measured as well, with precision improved by a factor of three compared to previous measurements. These $\chi_{c0}$ decay branching fractions provide important inputs for the study of glueballs., Comment: 9 pages, 1 figure

Software testing is a critical aspect of software development, yet generating test cases remains a routine task for engineers. This paper presents a benchmark, CLOVER, to evaluate models' capabilities in generating and completing test cases under specific conditions. Spanning from simple assertion completions to writing test cases that cover specific code blocks across multiple files, these tasks are based on 12 python repositories, analyzing 845 problems with context lengths ranging from 4k to 128k tokens. Utilizing code testing frameworks, we propose a method to construct retrieval contexts using coverage information. While models exhibit comparable performance with short contexts, notable differences emerge with 16k contexts. Notably, models like GPT-4o and Claude 3.5 can effectively leverage relevant snippets; however, all models score below 35\% on the complex Task III, even with the oracle context provided, underscoring the benchmark's significance and the potential for model improvement. The benchmark is containerized for code execution across tasks, and we will release the code, data, and construction methodologies., Comment: 16 pages

We propose a novel online Gaussian process (GP) model that is capable of capturing long-term memory in sequential data in an online regression setting. Our model, Online HiPPO Sparse Variational Gaussian Process Regression (OHSGPR), leverages the HiPPO (High-order Polynomial Projection Operators) framework, which is popularized in the RNN domain due to its long-range memory modeling capabilities. We interpret the HiPPO time-varying orthogonal projections as inducing variables with time-dependent orthogonal polynomial basis functions, which allows the SGPR inducing points to memorize the process history. We show that the HiPPO framework fits naturally into the interdomain GP framework and demonstrate that the kernel matrices can also be updated online in a recurrence form based on the ODE evolution of HiPPO. We evaluate our method on time series regression tasks, showing that it outperforms the existing online GP method in terms of predictive performance and computational efficiency, Comment: 13 pages, 4 figures

Referring Remote Sensing Image Segmentation (RRSIS) is critical for ecological monitoring, urban planning, and disaster management, requiring precise segmentation of objects in remote sensing imagery guided by textual descriptions. This task is uniquely challenging due to the considerable vision-language gap, the high spatial resolution and broad coverage of remote sensing imagery with diverse categories and small targets, and the presence of clustered, unclear targets with blurred edges. To tackle these issues, we propose \ours, a novel framework designed to bridge the vision-language gap, enhance multi-scale feature interaction, and improve fine-grained object differentiation. Specifically, \ours introduces: (1) the Bidirectional Spatial Correlation (BSC) for improved vision-language feature alignment, (2) the Target-Background TwinStream Decoder (T-BTD) for precise distinction between targets and non-targets, and (3) the Dual-Modal Object Learning Strategy (D-MOLS) for robust multimodal feature reconstruction. Extensive experiments on the benchmark datasets RefSegRS and RRSIS-D demonstrate that \ours achieves state-of-the-art performance. Specifically, \ours improves the overall IoU (oIoU) by 3.76 percentage points (80.57) and 1.44 percentage points (79.23) on the two datasets, respectively. Additionally, it outperforms previous methods in the mean IoU (mIoU) by 5.37 percentage points (67.95) and 1.84 percentage points (66.04), effectively addressing the core challenges of RRSIS with enhanced precision and robustness.

We report the position reconstruction methods and surface background model for the PandaX-4T dark matter direct search experiment. This work develops two position reconstruction algorithms: template matching (TM) method and photon acceptance function (PAF) method. Both methods determine the horizontal position of events based on the light pattern of secondary scintillation collected by the light sensors. After a comprehensive evaluation of resolution, uniformity, and robustness, the PAF method was selected for position reconstruction, while the TM method was employed for verification. The PAF method achieves a bulk event resolution of 1.0 mm and a surface event resolution of 4.4 mm for a typical $S2$ signal with a bottom charge of 1500 PE (about 14 keV). The uniformity is around 20\%. Robustness studies reveal average deviations of 5.1 mm and 8.8 mm for the commissioning run (Run0) and the first science run (Run1), respectively, due to the deactivation of certain PMTs. A data-driven surface background model is developed based on the PAF method. The surface background is estimated to be $0.09 \pm 0.06$ events for Run0 (0.54 tonne$\cdot$year) and $0.17 \pm 0.11$ events for Run1 (1.00 tonne$\cdot$year)., Comment: 22 pages, 15 figures, 2 tables

Fish use their lateral lines to sense flows and pressure gradients, enabling them to detect nearby objects and organisms. Towards replicating this capability, we demonstrated successful leader-follower formation swimming using flow pressure sensing in our undulatory robotic fish ($\mu$Bot/MUBot). The follower $\mu$Bot is equipped at its head with bilateral pressure sensors to detect signals excited by both its own and the leader's movements. First, using experiments with static formations between an undulating leader and a stationary follower, we determined the formation that resulted in strong pressure variations measured by the follower. This formation was then selected as the desired formation in free swimming for obtaining an expert policy. Next, a long short-term memory neural network was used as the control policy that maps the pressure signals along with the robot motor commands and the Euler angles (measured by the onboard IMU) to the steering command. The policy was trained to imitate the expert policy using behavior cloning and Dataset Aggregation (DAgger). The results show that with merely two bilateral pressure sensors and less than one hour of training data, the follower effectively tracked the leader within distances of up to 200 mm (= 1 body length) while swimming at speeds of 155 mm/s (= 0.8 body lengths/s). This work highlights the potential of fish-inspired robots to effectively navigate fluid environments and achieve formation swimming through the use of flow pressure feedback., Comment: 7 pages, 10 figures. Accepted for 2025 IEEE International Conference on Robotics and Automation (ICRA). Supplementary video: https://youtu.be/DIDYGi9Td0I

We study the $\alpha + \alpha + {}^{3}$He cluster structure of ${}^{11}$C within the microscopic cluster model. The calculations essentially reproduce the energy spectra for both negative and positive parity states, particularly the $3/2_3^-$ state near the $\alpha+\alpha$+${}^{3}$He threshold. We also calculate the isoscalar monopole, electric quadrupole transition strengths, and root-mean-square radii for the low-lying states. These results suggest that the $3/2_3^-$, $1/2_2^-$, and $5/2_3^-$ states have a well-developed $\alpha + \alpha$ + ${}^{3}$He cluster structure. The analysis of the generator coordinate method wave functions indicates the dilute gaslike nature for the $3/2_3^-$, $1/2_2^-$, and $5/2_3^-$ states, suggesting that they could be candidates for the Hoyle-analog state. Furthermore, it is found that the $5/2_2^+$ and $5/2_3^+$ states may possess a linear chain structure., Comment: 12 pages, 8 figures

The fractional shortcut to adiabaticity (f-STA) for production of quantum superposition states is proposed firstly via a three-level system with $\Lambda$-type linkage pattern and a four-level system with tripod structure. \textcolor[rgb]{1,0,0}{\sout{The fast, robust and efficient}} \textcolor[rgb]{0,0,1}{\uwave{The fast and robust}} production of the coherent superposition states is studied by comparing the populations for the f-STA and the fractional stimulated Raman adiabatic passage (f-STIRAP). The \textcolor[rgb]{1,0,0}{\sout{superposition}} states with equal proportion can be produced by fixing the controllable parameters of the driving pulses at the final moment of the whole process. The effects of the pulse intensity and the time delay of the pulses on the production process are discussed by monitoring the populations on all of the quantum states. In particular, the spontaneous emission arising from the intermediate state is investigated by the quantum master equation. The result reveals that the f-STA exhibits superior advantages over the f-STIRAP in producing the superposition states.

Unraveling the adsorption mechanism and thermodynamics of O$_2$ and H$_2$O on uranium dioxide surfaces is critical for the nuclear fuel storage and uranium corrosion. Based on the first-principles DFT+U-D3 calculations, we carefully test the effect of antiferromagnetic order arrangements on the thermodynamic stability of UO$_2$ surfaces and propose the 1k AFM surface computational model. The chemisorption states of O$_2$ and H$_2$O on UO$_2$ (111) surface, suggested by previous experiments, are accurately calculated for the first time. The adsorption properties of O$_2$ and H$_2$O on UO$_2$(111) and (110) surfaces are discussed in detail to reveal the different interaction mechanisms. Combined with ab initio atomistic thermodynamics method, we systematically calculate the chemisorption phase diagram and isotherm of O$_2$ and H$_2$O on UO$_2$ surfaces. Due to the different intermolecular interactions, the monolayer and multilayer adsorption models are identified for O$_2$ and H$_2$O, respectively. This study has comprehensively revealed the different adsorption mechanisms of O$_2$ and H$_2$O on UO$_2$ surfaces, bridging the electronic structure calculations to the interpretation of experimental results and providing a solid foundation for future theoretical studies of uranium corrosion mechanism in humid air.

In machine learning practice, early stopping has been widely used to regularize models and can save computational costs by halting the training process when the model's performance on a validation set stops improving. However, conventional early stopping applies the same stopping criterion to all instances without considering their individual learning statuses, which leads to redundant computations on instances that are already well-learned. To further improve the efficiency, we propose an Instance-dependent Early Stopping (IES) method that adapts the early stopping mechanism from the entire training set to the instance level, based on the core principle that once the model has mastered an instance, the training on it should stop. IES considers an instance as mastered if the second-order differences of its loss value remain within a small range around zero. This offers a more consistent measure of an instance's learning status compared with directly using the loss value, and thus allows for a unified threshold to determine when an instance can be excluded from further backpropagation. We show that excluding mastered instances from backpropagation can increase the gradient norms, thereby accelerating the decrease of the training loss and speeding up the training process. Extensive experiments on benchmarks demonstrate that IES method can reduce backpropagation instances by 10%-50% while maintaining or even slightly improving the test accuracy and transfer learning performance of a model., Comment: Accepted by ICLR 2025 (Spotlight)

Modern adaptive optimization methods, such as Adam and its variants, have emerged as the most widely used tools in deep learning over recent years. These algorithms offer automatic mechanisms for dynamically adjusting the update step based on estimates of gradient statistics. Compared to traditional algorithms like Stochastic Gradient Descent, these adaptive methods are typically more robust to model scale and hyperparameter tuning. However, the gradient statistics employed by these methods often do not leverage sufficient gradient covariance information, leading to suboptimal updates in certain directions of the parameter space and potentially slower convergence. In this work, we keep track of such covariance statistics in the form of a structured preconditioner matrix. Unlike other works, our approach does not apply direct approximations to estimate this matrix. We instead implement an invertible transformation that maps the preconditioner matrix into a new space where it becomes approximately diagonal. This enables a diagonal approximation of the preconditioner matrix in the transformed space, offering several computational advantages. Empirical results show that our approach can substantially enhance the convergence speed of modern adaptive optimizers. Notably, for large language models like LLaMA, we can achieve a speedup of 2x compared to the baseline Adam. Additionally, our method can be integrated with memory-efficient optimizers like Adafactor to manage computational overhead., Comment: 19 pages, 13 figures

We propose a novel mechanism for photon-dark photon mass state oscillations mediated by gravitational separation during propagation through the interstellar medium. This phenomenon establishes a new avenue for the detection of dark matter. By analyzing gravitational lensing data from quasars, we investigate the sensitivity of this approach to dark photons. Our analysis demonstrates constraints of$\epsilon<10^-2$ in the dark photon mass range of $10^{-14}eV$. Furthermore, we propose potential applications of this mechanism to astrophysical systems with strong gravitational fields, such as neutron stars and black hole accretion disks., Comment: 7 pages,3 figures

Using $e^+e^-$ collision data at 13 center-of-mass energies ranging from 4.600 to 4.950 GeV collected with the BESIII detector, we search for the unmeasured $e^+e^-\to K_S^0 K_S^0 h_c$ process . No significant signal is observed, and the upper limits of the Born cross sections at each center-of-mass energy are presented.

Nonreciprocal thermal emitters that break the conventional Kirchhoff's law allow independent control of emissivity and absorptivity and promise exciting new functionalities in controlling heat flow for thermal and energy applications. In enabling some of these applications, nonreciprocal thermal emitters will unavoidably need to serve as hot emitters. Leveraging magneto-optical effects, degenerate semiconductors have been demonstrated as a promising optical material platform for nonreciprocal thermal radiation. However, existing modeling and experimental efforts are limited to near room temperature (< 373 K), and it remains elusive whether nonreciprocal properties can persist at high temperatures. In this work, we demonstrate strong nonreciprocal radiative properties at temperatures up to 600 K. We propose a theoretical model by considering the temperature dependence of the key parameters for the nonreciprocal behavior and experimentally investigate the temperature dependence of the nonreciprocal properties of InAs, a degenerate semiconductor, using a customized angle-resolved high-temperature magnetic emissometry setup. Our theoretical model and experimental results show an agreement, revealing that strong nonreciprocity can persist at temperatures over 800 K for high-temperature stable semiconductors, enabling a pathway for nonreciprocal radiative heat flow control at high temperatures., Comment: 22 pages, 8 figures

Decentralized finance (DeFi) applications depend on accurate price oracles to ensure secure transactions, yet these oracles are highly vulnerable to manipulation, enabling attackers to exploit smart contract vulnerabilities for unfair asset valuation and financial gain. Detecting such manipulations traditionally relies on the manual effort of experienced experts, presenting significant challenges. In this paper, we propose a novel LLM-driven framework that automates the detection of price oracle manipulations by leveraging the complementary strengths of different LLM models (LLMs). Our approach begins with domain-specific knowledge extraction, where an LLM model synthesizes precise insights about price oracle vulnerabilities from top-tier academic papers, eliminating the need for profound expertise from developers or auditors. This knowledge forms the foundation for a second LLM model to generate structured, context-aware chain of thought prompts, which guide a third LLM model in accurately identifying manipulation patterns in smart contracts. We validate the effectiveness of framework through experiments on 60 known vulnerabilities from 46 real-world DeFi attacks or projects spanning 2021 to 2023. The best performing combination of LLMs (Haiku-Haiku-4o-mini) identified by AiRacleX demonstrate a 2.58-times improvement in recall (0.667 vs 0.259) compared to the state-of-the-art tool GPTScan, while maintaining comparable precision. Furthermore, our framework demonstrates the feasibility of replacing commercial models with open-source alternatives, enhancing privacy and security for developers.

Chip-scale integration of optical frequency combs, particularly soliton microcombs, enables miniaturized instrumentation for timekeeping, ranging, and spectroscopy. Although soliton microcombs have been demonstrated on various material platforms, realizing complete comb functionality on photonic chips requires the co-integration of high-speed modulators and efficient frequency doublers, features that are available in a monolithic form on X-cut thin-film lithium niobate (TFLN). However, the pronounced Raman nonlinearity associated with extraordinary light in this platform has so far precluded soliton microcomb generation. Here, we report the generation of transverse-electric-polarized soliton microcombs with a 25 GHz repetition rate in high-Q microresonators on X-cut TFLN chips. By precisely orienting the racetrack microresonator relative to the optical axis, we mitigate Raman nonlinearity and enable soliton formation under continuous-wave laser pumping. Moreover, the soliton microcomb spectra are extended to 350 nm with pulsed laser pumping. This work expands the capabilities of TFLN photonics and paves the way for the monolithic integration of fast-tunable, self-referenced microcombs.

In the case of the circular unitary ensemble of Haar distributed unitary matrices, the large $N$ form of the moments of the absolute value of the characteristic polynomial is known to be of relevance to the moments of the Riemann zeta function on the critical line, and further to relate to Fisher-Hartwig asymptotics in the theory of Toeplitz determinants. The constant (with respect to $N$) in this asymptotic expansion, involving the Barnes $G$ function, is known to appear in the analogous expansion for the moments of the characteristic polynomial in other classical ensembles with unitary symmetry, for example the Gaussian unitary ensemble. Using certain duality formulas for averages over powers of characteristic polynomials, this same effect is studied for classical $\beta$ ensembles, which is tractable in the case of even moments. For a given value of $\beta$, a distinguished factor in the asymptotic expansion can be identified which is universal with respect to the particular $\beta$ ensemble., Comment: 27 pages

We provide a new understanding of the stochastic gradient bandit algorithm by showing that it converges to a globally optimal policy almost surely using \emph{any} constant learning rate. This result demonstrates that the stochastic gradient algorithm continues to balance exploration and exploitation appropriately even in scenarios where standard smoothness and noise control assumptions break down. The proofs are based on novel findings about action sampling rates and the relationship between cumulative progress and noise, and extend the current understanding of how simple stochastic gradient methods behave in bandit settings., Comment: Updated version for a paper published at NeurIPS 2024

Recent advancements in large language models (LLMs) have significantly improved various natural language processing (NLP) tasks. Typically, LLMs are trained to predict the next token, aligning well with many NLP tasks. However, in knowledge graph (KG) scenarios, entities are the fundamental units and identifying an entity requires at least several tokens. This leads to a granularity mismatch between KGs and natural languages. To address this issue, we propose K-ON, which integrates KG knowledge into the LLM by employing multiple head layers for next k-step prediction. K-ON can not only generate entity-level results in one step, but also enables contrastive loss against entities, which is the most powerful tool in KG representation learning. Experimental results show that K-ON outperforms state-of-the-art methods that incorporate text and even the other modalities., Comment: AAAI 2025 (Oral)

LLMs are widely used for offensive language detection due to their advanced capability. However, the challenges posed by human annotation disagreement in real-world datasets remain underexplored. These disagreement samples are difficult to detect due to their ambiguous nature. Additionally, the confidence of LLMs in processing disagreement samples can provide valuable insights into their alignment with human annotators. To address this gap, we systematically evaluate the ability of LLMs to detect offensive language with annotation disagreement. We compare the binary accuracy of multiple LLMs across varying annotation agreement levels and analyze the relationship between LLM confidence and annotation agreement. Furthermore, we investigate the impact of disagreement samples on LLM decision-making during few-shot learning and instruction fine-tuning. Our findings highlight the challenges posed by disagreement samples and offer guidance for improving LLM-based offensive language detection., Comment: 17 pages, submitted to the ACL 2025

We identify two major limitations in the existing studies on retinal vessel segmentation: (1) Most existing works are restricted to one modality, i.e, the Color Fundus (CF). However, multi-modality retinal images are used every day in the study of retina and retinal diseases, and the study of vessel segmentation on the other modalities is scarce; (2) Even though a small amount of works extended their experiments to limited new modalities such as the Multi-Color Scanning Laser Ophthalmoscopy (MC), these works still require finetuning a separate model for the new modality. And the finetuning will require extra training data, which is difficult to acquire. In this work, we present a foundational universal vessel segmentation model (UVSM) for multi-modality retinal images. Not only do we perform the study on a much wider range of modalities, but also we propose a universal model to segment the vessels in all these commonly-used modalities. Despite being much more versatile comparing with existing methods, our universal model still demonstrates comparable performance with the state-of-the- art finetuned methods. To the best of our knowledge, this is the first work that achieves cross-modality retinal vessel segmentation and also the first work to study retinal vessel segmentation in some novel modalities.

We introduce a generalizable and unified framework to synthesize view-consistent and temporally coherent avatars from a single image, addressing the challenging problem of single-image avatar generation. While recent methods employ diffusion models conditioned on human templates like depth or normal maps, they often struggle to preserve appearance information due to the discrepancy between sparse driving signals and the actual human subject, resulting in multi-view and temporal inconsistencies. Our approach bridges this gap by combining the reconstruction power of regression-based 3D human reconstruction with the generative capabilities of a diffusion model. The dense driving signal from the initial reconstructed human provides comprehensive conditioning, ensuring high-quality synthesis faithful to the reference appearance and structure. Additionally, we propose a unified framework that enables the generalization learned from novel pose synthesis on in-the-wild videos to naturally transfer to novel view synthesis. Our video-based diffusion model enhances disentangled synthesis with high-quality view-consistent renderings for novel views and realistic non-rigid deformations in novel pose animation. Results demonstrate the superior generalization ability of our method across in-domain and out-of-domain in-the-wild datasets. Project page: https://humansensinglab.github.io/GAS/

Knowledge distillation (KD) is a powerful strategy for training deep neural networks (DNNs). Although it was originally proposed to train a more compact ``student'' model from a large ``teacher'' model, many recent efforts have focused on adapting it to promote generalization of the model itself, such as online KD and self KD. % as an effective way Here, we propose an accessible and compatible strategy named Spaced KD to improve the effectiveness of both online KD and self KD, in which the student model distills knowledge from a teacher model trained with a space interval ahead. This strategy is inspired by a prominent theory named \emph{spacing effect} in biological learning and memory, positing that appropriate intervals between learning trials can significantly enhance learning performance. With both theoretical and empirical analyses, we demonstrate that the benefits of the proposed Spaced KD stem from convergence to a flatter loss landscape during stochastic gradient descent (SGD). We perform extensive experiments to validate the effectiveness of Spaced KD in improving the learning performance of DNNs (e.g., the performance gain is up to 2.31\% and 3.34\% on Tiny-ImageNet over online KD and self KD, respectively).

Advanced Persistent Threats (APTs) are challenging to detect due to their complexity and stealth. To mitigate such attacks, many approaches utilize provenance graphs to model entities and their dependencies, detecting the covert and persistent nature of APTs. However, existing methods face several challenges: 1) Environmental noise hinders precise detection; 2) Reliance on hard-to-obtain labeled data and prior knowledge of APTs limits their ability to detect unknown threats; 3) The difficulty in capturing long-range interaction dependencies, leading to the loss of critical context. We propose Sentient, a threat detection system based on behavioral intent analysis that detects node-level threats from audit logs. Sentient constructs a provenance graph from the audit logs and uses this graph to build multiple scenarios. By combining graph comprehension with multiple scenario comprehension, Sentient learns normal interaction behaviors. Sentient detects anomalies by identifying interactions that deviate from the established behavior patterns. We evaluated Sentient on three widely used datasets covering both real-world and simulated attacks. The results confirm that Sentient consistently delivers strong detection performance. Notably, Sentient achieves entity-level APT detection with a precision of 96% and a recall of 99%.

Recent character image animation methods based on diffusion models, such as Animate Anyone, have made significant progress in generating consistent and generalizable character animations. However, these approaches fail to produce reasonable associations between characters and their environments. To address this limitation, we introduce Animate Anyone 2, aiming to animate characters with environment affordance. Beyond extracting motion signals from source video, we additionally capture environmental representations as conditional inputs. The environment is formulated as the region with the exclusion of characters and our model generates characters to populate these regions while maintaining coherence with the environmental context. We propose a shape-agnostic mask strategy that more effectively characterizes the relationship between character and environment. Furthermore, to enhance the fidelity of object interactions, we leverage an object guider to extract features of interacting objects and employ spatial blending for feature injection. We also introduce a pose modulation strategy that enables the model to handle more diverse motion patterns. Experimental results demonstrate the superior performance of the proposed method., Comment: Project Page: https://humanaigc.github.io/animate-anyone-2/