Search

Kilonova is an optical-infrared transient powered by the radioactive decay of heavy nuclei from binary neutron star mergers. Its observational characteristics depend on the mass and the nuclide composition of meger ejecta, which are sensitive to the equation of state (EoS) of neutron star. We use astrophysical conditions derived from different EoSs as nucleosynthesis inputs to explore the impact of various EoS on the $r$-process nucleosynthesis and the kilonova emission. Our results show that both the abundance patterns of merger ejecta and kilonova light curves are strongly dependent on the neutron star EoSs. Given the mass of two neutron stars, the merger with a softer EoS tends to generate a larger amount of ejected material, and may lead to a brighter kilonova peak luminosity. The relationship between the neutron star EoS and the peak luminosity provides a probe for constraining the properties of EoS in multi-messenger observations of neutron star mergers., Comment: 9 pages, 6 figures. Accepted for publication in RAA

Recent advancements in Large Language Models (LLMs) have created new opportunities to enhance performance on complex reasoning tasks by leveraging test-time computation. However, conventional approaches such as repeated sampling with majority voting or reward model scoring, often face diminishing returns as test-time compute scales, in addition to requiring costly task-specific reward model training. In this paper, we present Self-Enhanced Test-Time Scaling (SETS), a novel method that leverages the self-verification and self-correction capabilities of recent advanced LLMs to overcome these limitations. SETS integrates sampling, self-verification, and self-correction into a unified framework, enabling efficient and scalable test-time computation for improved capabilities at complex tasks. Through extensive experiments on challenging planning and reasoning benchmarks, compared to the alternatives, we demonstrate that SETS achieves significant performance improvements and more favorable test-time scaling laws.

This paper addresses the challenges in learning-based monocular positioning by proposing VKFPos, a novel approach that integrates Absolute Pose Regression (APR) and Relative Pose Regression (RPR) via an Extended Kalman Filter (EKF) within a variational Bayesian inference framework. Our method shows that the essential posterior probability of the monocular positioning problem can be decomposed into APR and RPR components. This decomposition is embedded in the deep learning model by predicting covariances in both APR and RPR branches, allowing them to account for associated uncertainties. These covariances enhance the loss functions and facilitate EKF integration. Experimental evaluations on both indoor and outdoor datasets show that the single-shot APR branch achieves accuracy on par with state-of-the-art methods. Furthermore, for temporal positioning, where consecutive images allow for RPR and EKF integration, VKFPos outperforms temporal APR and model-based integration methods, achieving superior accuracy.

This paper presents SANA-1.5, a linear Diffusion Transformer for efficient scaling in text-to-image generation. Building upon SANA-1.0, we introduce three key innovations: (1) Efficient Training Scaling: A depth-growth paradigm that enables scaling from 1.6B to 4.8B parameters with significantly reduced computational resources, combined with a memory-efficient 8-bit optimizer. (2) Model Depth Pruning: A block importance analysis technique for efficient model compression to arbitrary sizes with minimal quality loss. (3) Inference-time Scaling: A repeated sampling strategy that trades computation for model capacity, enabling smaller models to match larger model quality at inference time. Through these strategies, SANA-1.5 achieves a text-image alignment score of 0.72 on GenEval, which can be further improved to 0.80 through inference scaling, establishing a new SoTA on GenEval benchmark. These innovations enable efficient model scaling across different compute budgets while maintaining high quality, making high-quality image generation more accessible.

Zero-knowledge proofs (ZKPs) are widely applied in digital economies, such as cryptocurrencies and smart contracts, for establishing trust and ensuring privacy between untrusted parties. However, almost all ZKPs rely on unproven computational assumptions or are vulnerable to quantum adversaries. We propose and experimentally implement an unconditionally secure ZKP for the graph three-coloring problem by combining subset relativistic bit commitments with quantum nonlocality game. Our protocol achieves a linear relationship between interactive rounds and the number of edges, reducing round complexity and storage requirements by thirteen orders of magnitude, thereby significantly enhancing practical feasibility. Our work illustrates the powerful potential of integrating special relativity with quantum theory in trustless cryptography, paving the way for robust applications against quantum attacks in distrustful internet environments., Comment: 24 pages, 8 figures

Recent advancements in Vision-Language-Action (VLA) models have leveraged pre-trained Vision-Language Models (VLMs) to improve the generalization capabilities. VLMs, typically pre-trained on vision-language understanding tasks, provide rich semantic knowledge and reasoning abilities. However, prior research has shown that VLMs often focus on high-level semantic content and neglect low-level features, limiting their ability to capture detailed spatial information and understand physical dynamics. These aspects, which are crucial for embodied control tasks, remain underexplored in existing pre-training paradigms. In this paper, we investigate the training paradigm for VLAs, and introduce \textbf{UP-VLA}, a \textbf{U}nified VLA model training with both multi-modal \textbf{U}nderstanding and future \textbf{P}rediction objectives, enhancing both high-level semantic comprehension and low-level spatial understanding. Experimental results show that UP-VLA achieves a 33% improvement on the Calvin ABC-D benchmark compared to the previous state-of-the-art method. Additionally, UP-VLA demonstrates improved success rates in real-world manipulation tasks, particularly those requiring precise spatial information.

Magnetic Resonance Imaging (MRI) offers critical insights into microstructural details, however, the spatial resolution of standard 1.5T imaging systems is often limited. In contrast, 7T MRI provides significantly enhanced spatial resolution, enabling finer visualization of anatomical structures. Though this, the high cost and limited availability of 7T MRI hinder its widespread use in clinical settings. To address this challenge, a novel Super-Resolution (SR) model is proposed to generate 7T-like MRI from standard 1.5T MRI scans. Our approach leverages a diffusion-based architecture, incorporating gradient nonlinearity correction and bias field correction data from 7T imaging as guidance. Moreover, to improve deployability, a progressive distillation strategy is introduced. Specifically, the student model refines the 7T SR task with steps, leveraging feature maps from the inference phase of the teacher model as guidance, aiming to allow the student model to achieve progressively 7T SR performance with a smaller, deployable model size. Experimental results demonstrate that our baseline teacher model achieves state-of-the-art SR performance. The student model, while lightweight, sacrifices minimal performance. Furthermore, the student model is capable of accepting MRI inputs at varying resolutions without the need for retraining, significantly further enhancing deployment flexibility. The clinical relevance of our proposed method is validated using clinical data from Massachusetts General Hospital. Our code is available at https://github.com/ZWang78/SR.

Low-rank Adaptation (LoRA) has demonstrated remarkable capabilities for task specific fine-tuning. However, in scenarios that involve multiple tasks, training a separate LoRA model for each one results in considerable inefficiency in terms of storage and inference. Moreover, existing parameter generation methods fail to capture the correlations among these tasks, making multi-task LoRA parameter generation challenging. To address these limitations, we propose In-Context Meta LoRA (ICM-LoRA), a novel approach that efficiently achieves task-specific customization of large language models (LLMs). Specifically, we use training data from all tasks to train a tailored generator, Conditional Variational Autoencoder (CVAE). CVAE takes task descriptions as inputs and produces task-aware LoRA weights as outputs. These LoRA weights are then merged with LLMs to create task-specialized models without the need for additional fine-tuning. Furthermore, we utilize in-context meta-learning for knowledge enhancement and task mapping, to capture the relationship between tasks and parameter distributions. As a result, our method achieves more accurate LoRA parameter generation for diverse tasks using CVAE. ICM-LoRA enables more accurate LoRA parameter reconstruction than current parameter reconstruction methods and is useful for implementing task-specific enhancements of LoRA parameters. At the same time, our method occupies 283MB, only 1\% storage compared with the original LoRA.

Machine-learning models are increasingly deployed on resource-constrained embedded systems with strict timing constraints. In such scenarios, the worst-case execution time (WCET) of the models is required to ensure safe operation. Specifically, decision trees are a prominent class of machine-learning models and the main building blocks of tree-based ensemble models (e.g., random forests), which are commonly employed in resource-constrained embedded systems. In this paper, we develop a systematic approach for WCET optimization of decision tree implementations. To this end, we introduce a linear surrogate model that estimates the execution time of individual paths through a decision tree based on the path's length and the number of taken branches. We provide an optimization algorithm that constructively builds a WCET-optimal implementation of a given decision tree with respect to this surrogate model. We experimentally evaluate both the surrogate model and the WCET-optimization algorithm. The evaluation shows that the optimization algorithm improves analytically determined WCET by up to $17\%$ compared to an unoptimized implementation.

Real-time ego-motion tracking for endoscope is a significant task for efficient navigation and robotic automation of endoscopy. In this paper, a novel framework is proposed to perform real-time ego-motion tracking for endoscope. Firstly, a multi-modal visual feature learning network is proposed to perform relative pose prediction, in which the motion feature from the optical flow, the scene features and the joint feature from two adjacent observations are all extracted for prediction. Due to more correlation information in the channel dimension of the concatenated image, a novel feature extractor is designed based on an attention mechanism to integrate multi-dimensional information from the concatenation of two continuous frames. To extract more complete feature representation from the fused features, a novel pose decoder is proposed to predict the pose transformation from the concatenated feature map at the end of the framework. At last, the absolute pose of endoscope is calculated based on relative poses. The experiment is conducted on three datasets of various endoscopic scenes and the results demonstrate that the proposed method outperforms state-of-the-art methods. Besides, the inference speed of the proposed method is over 30 frames per second, which meets the real-time requirement. The project page is here: \href{https://remote-bmxs.netlify.app}{remote-bmxs.netlify.app}

Conversational human-AI interaction (CHAI) have recently driven mainstream adoption of AI. However, CHAI poses two key challenges for designers and researchers: users frequently have ambiguous goals and an incomplete understanding of AI functionalities, and the interactions are brief and transient, limiting opportunities for sustained engagement with users. AI agents can help address these challenges by suggesting contextually relevant prompts, by standing in for users during early design testing, and by helping users better articulate their goals. Guided by research-through-design, we explored agentic AI workflows through the development and testing of a probe over four iterations with 10 users. We present our findings through an annotated portfolio of design artifacts, and through thematic analysis of user experiences, offering solutions to the problems of ambiguity and transient in CHAI. Furthermore, we examine the limitations and possibilities of these AI agent workflows, suggesting that similar collaborative approaches between humans and AI could benefit other areas of design.

Current medical AI systems often fail to replicate real-world clinical reasoning, as they are predominantly trained and evaluated on static text and question-answer tasks. These tuning methods and benchmarks overlook critical aspects like evidence-based reasoning and handling distracting information. To bridge this gap, we introduce a novel benchmark that simulates real-world diagnostic scenarios, integrating noise and difficulty levels aligned with USMLE standards. Moreover, we explore dialogue-based fine-tuning, which transforms static datasets into conversational formats to better capture iterative reasoning processes. Experiments show that dialogue-tuned models outperform traditional methods, with improvements of $9.64\%$ in multi-round reasoning scenarios and $6.18\%$ in accuracy in a noisy environment. Our findings highlight dialogue tuning as a promising approach for advancing clinically aligned and robust medical AI systems.

We present BreezyVoice, a Text-to-Speech (TTS) system specifically adapted for Taiwanese Mandarin, highlighting phonetic control abilities to address the unique challenges of polyphone disambiguation in the language. Building upon CosyVoice, we incorporate a $S^{3}$ tokenizer, a large language model (LLM), an optimal-transport conditional flow matching model (OT-CFM), and a grapheme to phoneme prediction model, to generate realistic speech that closely mimics human utterances. Our evaluation demonstrates BreezyVoice's superior performance in both general and code-switching contexts, highlighting its robustness and effectiveness in generating high-fidelity speech. Additionally, we address the challenges of generalizability in modeling long-tail speakers and polyphone disambiguation. Our approach significantly enhances performance and offers valuable insights into the workings of neural codec TTS systems.

This paper presents Contourformer, a real-time contour-based instance segmentation algorithm. The method is fully based on the DETR paradigm and achieves end-to-end inference through iterative and progressive mechanisms to optimize contours. To improve efficiency and accuracy, we develop two novel techniques: sub-contour decoupling mechanisms and contour fine-grained distribution refinement. In the sub-contour decoupling mechanism, we propose a deformable attention-based module that adaptively selects sampling regions based on the current predicted contour, enabling more effective capturing of object boundary information. Additionally, we design a multi-stage optimization process to enhance segmentation precision by progressively refining sub-contours. The contour fine-grained distribution refinement technique aims to further improve the ability to express fine details of contours. These innovations enable Contourformer to achieve stable and precise segmentation for each instance while maintaining real-time performance. Extensive experiments demonstrate the superior performance of Contourformer on multiple benchmark datasets, including SBD, COCO, and KINS. We conduct comprehensive evaluations and comparisons with existing state-of-the-art methods, showing significant improvements in both accuracy and inference speed. This work provides a new solution for contour-based instance segmentation tasks and lays a foundation for future research, with the potential to become a strong baseline method in this field.

The state-of-the-art online learning models generally conduct a single online gradient descent when a new sample arrives and thus suffer from suboptimal model weights. To this end, we introduce an online broad learning system framework with closed-form solutions for each online update. Different from employing existing incremental broad learning algorithms for online learning tasks, which tend to incur degraded accuracy and expensive online update overhead, we design an effective weight estimation algorithm and an efficient online updating strategy to remedy the above two deficiencies, respectively. Specifically, an effective weight estimation algorithm is first developed by replacing notorious matrix inverse operations with Cholesky decomposition and forward-backward substitution to improve model accuracy. Second, we devise an efficient online updating strategy that dramatically reduces online update time. Theoretical analysis exhibits the splendid error bound and low time complexity of our model. The most popular test-then-training evaluation experiments on various real-world datasets prove its superiority and efficiency. Furthermore, our framework is naturally extended to data stream scenarios with concept drift and exceeds state-of-the-art baselines.

Utilizing the phenomenological theory based on crystal symmetry operation, we have established the complete angular dependencies of the galvanomagnetic effects, encompassing both anisotropic magnetoresistance (AMR) and the planar Hall effect (PHE), for the ferromagnetic films with C4v symmetry. These dependencies were experimentally confirmed via comprehensive angular-mapping of AMR and PHE in single-crystal Fe(001) films at room temperature. We demonstrated that the intrinsic magnetization-induced effects are independent of the field strength by carefully separating the field-induced and magnetization-induced galvanomagnetic effects. Our theoretical and experimental findings highlight the absence of in-plane four-fold angular dependence in PHE, a feature prohibited by the Onsager relation in systems with C4 symmetry. This study affirms that the universal angular dependencies of AMR and PHE in single crystals can be accurately predicted by the conventional phenomenological theory., Comment: 32 pages, 6 figures, with supplemental material

Recent studies have successfully integrated large vision-language models (VLMs) into low-level robotic control by supervised fine-tuning (SFT) with expert robotic datasets, resulting in what we term vision-language-action (VLA) models. Although the VLA models are powerful, how to improve these large models during interaction with environments remains an open question. In this paper, we explore how to further improve these VLA models via Reinforcement Learning (RL), a commonly used fine-tuning technique for large models. However, we find that directly applying online RL to large VLA models presents significant challenges, including training instability that severely impacts the performance of large models, and computing burdens that exceed the capabilities of most local machines. To address these challenges, we propose iRe-VLA framework, which iterates between Reinforcement Learning and Supervised Learning to effectively improve VLA models, leveraging the exploratory benefits of RL while maintaining the stability of supervised learning. Experiments in two simulated benchmarks and a real-world manipulation suite validate the effectiveness of our method., Comment: Accepted to ICRA 2025

The emergence of multimodal large language models (MLLMs) advances multimodal emotion recognition (MER) to the next level-from naive discriminative tasks to complex emotion understanding with advanced video understanding abilities and natural language description. However, the current community suffers from a lack of large-scale datasets with intensive, descriptive emotion annotations, as well as a multimodal-centric framework to maximize the potential of MLLMs for emotion understanding. To address this, we establish a new benchmark for MLLM-based emotion understanding with a novel dataset (MER-Caption), and a new model (AffectGPT). Utilizing our model-based crowd-sourcing data collection strategy, we construct the largest descriptive emotion dataset to date (by far), featuring over 2K fine-grained emotion categories across 115K samples. We also introduce the AffectGPT model, designed with pre-fusion operations to enhance multimodal integration. Finally, we present MER-UniBench, a unified benchmark with evaluation metrics tailored for both typical MER tasks and the free-form, natural language output style of MLLMs. Extensive experimental results demonstrate AffectGPT's robust performance across various MER tasks. We are publicly releasing both the AffectGPT model and the MER-Caption dataset to foster further research and development in emotion understanding.

Dynamic functional connectivity (dFC) using resting-state functional magnetic resonance imaging (rs-fMRI) is an advanced technique for capturing the dynamic changes of neural activities, and can be very useful in the studies of brain diseases such as Alzheimer's disease (AD). Yet, existing studies have not fully leveraged the sequential information embedded within dFC that can potentially provide valuable information when identifying brain conditions. In this paper, we propose a novel framework that jointly learns the embedding of both spatial and temporal information within dFC based on the transformer architecture. Specifically, we first construct dFC networks from rs-fMRI data through a sliding window strategy. Then, we simultaneously employ a temporal block and a spatial block to capture higher-order representations of dynamic spatio-temporal dependencies, via mapping them into an efficient fused feature representation. To further enhance the robustness of these feature representations by reducing the dependency on labeled data, we also introduce a contrastive learning strategy to manipulate different brain states. Experimental results on 345 subjects with 570 scans from the Alzheimer's Disease Neuroimaging Initiative (ADNI) demonstrate the superiority of our proposed method for MCI (Mild Cognitive Impairment, the prodromal stage of AD) prediction, highlighting its potential for early identification of AD.

Understanding brain disorders is crucial for accurate clinical diagnosis and treatment. Recent advances in Multimodal Large Language Models (MLLMs) offer a promising approach to interpreting medical images with the support of text descriptions. However, previous research has primarily focused on 2D medical images, leaving richer spatial information of 3D images under-explored, and single-modality-based methods are limited by overlooking the critical clinical information contained in other modalities. To address this issue, this paper proposes Brain-Adapter, a novel approach that incorporates an extra bottleneck layer to learn new knowledge and instill it into the original pre-trained knowledge. The major idea is to incorporate a lightweight bottleneck layer to train fewer parameters while capturing essential information and utilize a Contrastive Language-Image Pre-training (CLIP) strategy to align multimodal data within a unified representation space. Extensive experiments demonstrated the effectiveness of our approach in integrating multimodal data to significantly improve the diagnosis accuracy without high computational costs, highlighting the potential to enhance real-world diagnostic workflows.

Large language models (LLMs) exhibit remarkable capabilities in visual inspection of medical time-series data, achieving proficiency comparable to human clinicians. However, their broad scope limits domain-specific precision, and proprietary weights hinder fine-tuning for specialized datasets. In contrast, small specialized models (SSMs) excel in targeted tasks but lack the contextual reasoning required for complex clinical decision-making. To address these challenges, we propose ConMIL (Conformalized Multiple Instance Learning), a decision-support SSM that integrates seamlessly with LLMs. By using Multiple Instance Learning (MIL) to identify clinically significant signal segments and conformal prediction for calibrated set-valued outputs, ConMIL enhances LLMs' interpretative capabilities for medical time-series analysis. Experimental results demonstrate that ConMIL significantly improves the performance of state-of-the-art LLMs, such as ChatGPT4.0 and Qwen2-VL-7B. Specifically, \ConMIL{}-supported Qwen2-VL-7B achieves 94.92% and 96.82% precision for confident samples in arrhythmia detection and sleep staging, compared to standalone LLM accuracy of 46.13% and 13.16%. These findings highlight the potential of ConMIL to bridge task-specific precision and broader contextual reasoning, enabling more reliable and interpretable AI-driven clinical decision support.

An ideal multimodal agent should be aware of the quality of its input modalities. Recent advances have enabled large language models (LLMs) to incorporate auditory systems for handling various speech-related tasks. However, most audio LLMs remain unaware of the quality of the speech they process. This limitation arises because speech quality evaluation is typically excluded from multi-task training due to the lack of suitable datasets. To address this, we introduce the first natural language-based speech evaluation corpus, generated from authentic human ratings. In addition to the overall Mean Opinion Score (MOS), this corpus offers detailed analysis across multiple dimensions and identifies causes of quality degradation. It also enables descriptive comparisons between two speech samples (A/B tests) with human-like judgment. Leveraging this corpus, we propose an alignment approach with LLM distillation (ALLD) to guide the audio LLM in extracting relevant information from raw speech and generating meaningful responses. Experimental results demonstrate that ALLD outperforms the previous state-of-the-art regression model in MOS prediction, with a mean square error of 0.17 and an A/B test accuracy of 98.6%. Additionally, the generated responses achieve BLEU scores of 25.8 and 30.2 on two tasks, surpassing the capabilities of task-specific models. This work advances the comprehensive perception of speech signals by audio LLMs, contributing to the development of real-world auditory and sensory intelligent agents., Comment: ICLR 2025

It has long been a problem to arrange and execute irregular workloads on massively parallel devices. We propose a general framework for statically batching irregular workloads into a single kernel with a runtime task mapping mechanism on GPUs. We further apply this framework to Mixture-of-Experts (MoE) model inference and implement an optimized and efficient CUDA kernel. Our MoE kernel achieves up to 91% of the peak Tensor Core throughput on NVIDIA H800 GPU and 95% on NVIDIA H20 GPU., Comment: 11 pages

Long-tailed learning has garnered increasing attention due to its practical significance. Among the various approaches, the fine-tuning paradigm has gained considerable interest with the advent of foundation models. However, most existing methods primarily focus on leveraging knowledge from these models, overlooking the inherent biases introduced by the imbalanced training data they rely on. In this paper, we examine how such imbalances from pre-training affect long-tailed downstream tasks. Specifically, we find the imbalance biases inherited in foundation models on downstream task as parameter imbalance and data imbalance. During fine-tuning, we observe that parameter imbalance plays a more critical role, while data imbalance can be mitigated using existing re-balancing strategies. Moreover, we find that parameter imbalance cannot be effectively addressed by current re-balancing techniques, such as adjusting the logits, during training, unlike data imbalance. To tackle both imbalances simultaneously, we build our method on causal learning and view the incomplete semantic factor as the confounder, which brings spurious correlations between input samples and labels. To resolve the negative effects of this, we propose a novel backdoor adjustment method that learns the true causal effect between input samples and labels, rather than merely fitting the correlations in the data. Notably, we achieve an average performance increase of about $1.67\%$ on each dataset.

Recent advancements in speech generation have been driven by the large-scale training datasets. However, current models fall short of capturing the spontaneity and variability inherent in real-world human speech, due to their reliance on audiobook datasets limited to formal read-aloud speech styles. To bridge this gap, we introduce Emilia-Pipe, an open-source preprocessing pipeline to extract high-quality training data from valuable yet underexplored in-the-wild data that capture spontaneous human speech in real-world contexts. By leveraging Emilia-Pipe, we construct Emilia, the first multilingual speech generation dataset derived from in-the-wild speech data. This dataset comprises over 101k hours of speech across six languages: English, Chinese, German, French, Japanese, and Korean. Besides, we expand Emilia to Emilia-Large, a dataset exceeding 216k hours, making it the largest open-source speech generation dataset available. Extensive experiments demonstrate that Emilia significantly outperforms traditional audiobook datasets in generating spontaneous and human-like speech, showcasing superior performance in capturing diverse speaker timbre and speaking styles of real-world human speech. Furthermore, this work underscores the importance of scaling dataset size to advance speech generation research and validates the effectiveness of Emilia for both multilingual and crosslingual speech generation., Comment: Extended version of arXiv:2407.05361, submitted to TASLP, dataset is available at: https://huggingface.co/datasets/amphion/Emilia-Dataset

Feature modeling, which involves feature representation learning and leveraging, plays an essential role in industrial recommendation systems. However, the data distribution in real-world applications usually follows a highly skewed long-tail pattern due to the popularity bias, which easily leads to over-reliance on ID-based features, such as user/item IDs and ID sequences of interactions. Such over-reliance makes it hard for models to learn features comprehensively, especially for those non-ID meta features, e.g., user/item characteristics. Further, it limits the feature leveraging ability in models, getting less generalized and more susceptible to data noise. Previous studies on feature modeling focus on feature extraction and interaction, hardly noticing the problems brought about by the long-tail data distribution. To achieve better feature representation learning and leveraging on real-world data, we propose a model-agnostic framework AdaF^2M^2, short for Adaptive Feature Modeling with Feature Mask. The feature-mask mechanism helps comprehensive feature learning via multi-forward training with augmented samples, while the adapter applies adaptive weights on features responsive to different user/item states. By arming base models with AdaF^2M^2, we conduct online A/B tests on multiple recommendation scenarios, obtaining +1.37% and +1.89% cumulative improvements on user active days and app duration respectively. Besides, the extended offline experiments on different models show improvements as well. AdaF$^2$M$^2$ has been widely deployed on both retrieval and ranking tasks in multiple applications of Douyin Group, indicating its superior effectiveness and universality., Comment: Accepted by DASFAA2025

In industrial settings, the accurate detection of anomalies is essential for maintaining product quality and ensuring operational safety. Traditional industrial anomaly detection (IAD) models often struggle with flexibility and adaptability, especially in dynamic production environments where new defect types and operational changes frequently arise. Recent advancements in Multimodal Large Language Models (MLLMs) hold promise for overcoming these limitations by combining visual and textual information processing capabilities. MLLMs excel in general visual understanding due to their training on large, diverse datasets, but they lack domain-specific knowledge, such as industry-specific defect tolerance levels, which limits their effectiveness in IAD tasks. To address these challenges, we propose Echo, a novel multi-expert framework designed to enhance MLLM performance for IAD. Echo integrates four expert modules: Reference Extractor which provides a contextual baseline by retrieving similar normal images, Knowledge Guide which supplies domain-specific insights, Reasoning Expert which enables structured, stepwise reasoning for complex queries, and Decision Maker which synthesizes information from all modules to deliver precise, context-aware responses. Evaluated on the MMAD benchmark, Echo demonstrates significant improvements in adaptability, precision, and robustness, moving closer to meeting the demands of real-world industrial anomaly detection., Comment: 16 pages, 11 figures

Trustworthiness is essential for the precise and interpretable application of artificial intelligence (AI) in medical imaging. Traditionally, precision and interpretability have been addressed as separate tasks, namely medical image analysis and explainable AI, each developing its own models independently. In this study, for the first time, we investigate the development of a unified medical vision-language pre-training model that can achieve both accurate analysis and interpretable understanding of medical images across various modalities. To build the model, we construct MedConcept-23M, a large-scale dataset comprising 23 million medical image-text pairs extracted from 6.2 million scientific articles, enriched with concepts from the Unified Medical Language System (UMLS). Based on MedConcept-23M, we introduce ConceptCLIP, a medical AI model utilizing concept-enhanced contrastive language-image pre-training. The pre-training of ConceptCLIP involves two primary components: image-text alignment learning (IT-Align) and patch-concept alignment learning (PC-Align). This dual alignment strategy enhances the model's capability to associate specific image regions with relevant concepts, thereby improving both the precision of analysis and the interpretability of the AI system. We conducted extensive experiments on 5 diverse types of medical image analysis tasks, spanning 51 subtasks across 10 image modalities, with the broadest range of downstream tasks. The results demonstrate the effectiveness of the proposed vision-language pre-training model. Further explainability analysis across 6 modalities reveals that ConceptCLIP achieves superior performance, underscoring its robust ability to advance explainable AI in medical imaging. These findings highlight ConceptCLIP's capability in promoting trustworthy AI in the field of medicine.

Multimodal large language models (MLLMs) have shown impressive capabilities across various domains, excelling in processing and understanding information from multiple modalities. Despite the rapid progress made previously, insufficient OCR ability hinders MLLMs from excelling in text-related tasks. In this paper, we present \textbf{Ocean-OCR}, a 3B MLLM with state-of-the-art performance on various OCR scenarios and comparable understanding ability on general tasks. We employ Native Resolution ViT to enable variable resolution input and utilize a substantial collection of high-quality OCR datasets to enhance the model performance. We demonstrate the superiority of Ocean-OCR through comprehensive experiments on open-source OCR benchmarks and across various OCR scenarios. These scenarios encompass document understanding, scene text recognition, and handwritten recognition, highlighting the robust OCR capabilities of Ocean-OCR. Note that Ocean-OCR is the first MLLM to outperform professional OCR models such as TextIn and PaddleOCR.

Electrides as a unique class of emerging materials exhibit fascinating properties and hold important significance for understanding the matter under extreme conditions, which is characterized by valence electrons localized into the interstitial space as quasi-atoms (ISQs). In this work, using crystal structure prediction and first-principles calculations, we identified seven stable phases of Mg-Li that are electride with novel electronic properties under high pressure. Among them, MgLi10 is a semiconductor with a band gap of 0.22 eV; and Pm-3m MgLi is superconductor with a superconducting transition temperature of 22.8 K. The important role played by the localization degree of ISQ in the superconducting transition temperature of these electrides is revealed by systematic comparison of Mg-Li with other Li-rich electride superconductors. Furthermore, we proved that Pm-3m MgLi and Pnma MgLi also have distinct topological behavior with metallic surface states and the non-zero $Z_2$ invariant. The simultaneous coexistence of superconductivity, electronic band topology and electride property in the same structure of Pm-3m MgLi and Pnma MgLi demonstrates the feasibility of realizing multi-quantum phases in a single material, which will stimulate further research in these interdisciplinary fields., Comment: 38 pages, 7 figures, with Supporting Information

Using $\psi(3686)\rightarrow \pi^{0} h_{c}$ decays from a data sample of $(27.12\pm0.14)\times10^{8}$ $\psi(3686)$ events collected by the BESIII detector at the BEPCII collider, $h_c$ radiative decays to $\gamma\pi^{+}\pi^{-},~\gamma\pi^{+}\pi^{-}\eta,~\gamma2(\pi^{+}\pi^{-})$, and $\gamma p\bar{p}$ are observed for the first time, each with a significance greater than $5\sigma$. The corresponding branching fractions are measured. Furthermore, intermediate states below 2.8 GeV/$c^{2}$ are investigated, leading to the first observation of the decay process of $h_c\rightarrow\gamma f_{2}(1270)\rightarrow\gamma\pi^{+}\pi^{-}$ with a significance of $5.5\,\sigma$. This observation represents the first instance of $h_c$ radiative decay to a tensor state.

Classifier-Free Guidance (CFG) has been a default technique in various visual generative models, yet it requires inference from both conditional and unconditional models during sampling. We propose to build visual models that are free from guided sampling. The resulting algorithm, Guidance-Free Training (GFT), matches the performance of CFG while reducing sampling to a single model, halving the computational cost. Unlike previous distillation-based approaches that rely on pretrained CFG networks, GFT enables training directly from scratch. GFT is simple to implement. It retains the same maximum likelihood objective as CFG and differs mainly in the parameterization of conditional models. Implementing GFT requires only minimal modifications to existing codebases, as most design choices and hyperparameters are directly inherited from CFG. Our extensive experiments across five distinct visual models demonstrate the effectiveness and versatility of GFT. Across domains of diffusion, autoregressive, and masked-prediction modeling, GFT consistently achieves comparable or even lower FID scores, with similar diversity-fidelity trade-offs compared with CFG baselines, all while being guidance-free. Code will be available at https://github.com/thu-ml/GFT.

Traditional wireless image transmission methods struggle to balance rate efficiency and reconstruction quality under varying channel conditions. To address these challenges, we propose a novel semantic communication (SemCom) system that integrates entropy-aware and channel-adaptive mechanisms for wireless image transmission over multi-user multiple-input multiple-output (MU-MIMO) fading channels. Unlike existing approaches, our system dynamically adjusts transmission rates based on the entropy of feature maps, channel state information (CSI), and signal-to-noise ratio (SNR), ensuring optimal resource utilization and robust performance. The system employs feature map pruning, channel attention, spatial attention, and multihead self-attention (MHSA) mechanisms to prioritize critical semantic features and effectively reconstruct images. Experimental results demonstrate that the proposed system outperforms state-of-the-art benchmarks, including BPG+LDPC+4QAM and Deep JSCC, in terms of rate-distortion performance, flexibility, and robustness, particularly under challenging conditions such as low SNR, imperfect CSI, and inter-user interference. This work establishes a strong foundation for adaptive-rate SemCom systems and highlights their potential for real-time, bandwidthintensive applications.

Procrastination, the voluntary delay of tasks despite potential negative consequences, has prompted numerous time and task management interventions in the HCI community. While these interventions have shown promise in addressing specific behaviors, psychological theories suggest that learning about procrastination itself may help individuals develop their own coping strategies and build mental resilience. However, little research has explored how to support this learning process through HCI approaches. We present ProcrastiMate, a text adventure game where players learn about procrastination's causes and experiment with coping strategies by guiding in-game characters in managing relatable scenarios. Our field study with 27 participants revealed that ProcrastiMate facilitated learning and self-reflection while maintaining psychological distance, motivating players to integrate newly acquired knowledge in daily life. This paper contributes empirical insights on leveraging serious games to facilitate learning about procrastination and offers design implications for addressing psychological challenges through HCI approaches.

We introduce Baichuan-Omni-1.5, an omni-modal model that not only has omni-modal understanding capabilities but also provides end-to-end audio generation capabilities. To achieve fluent and high-quality interaction across modalities without compromising the capabilities of any modality, we prioritized optimizing three key aspects. First, we establish a comprehensive data cleaning and synthesis pipeline for multimodal data, obtaining about 500B high-quality data (text, audio, and vision). Second, an audio-tokenizer (Baichuan-Audio-Tokenizer) has been designed to capture both semantic and acoustic information from audio, enabling seamless integration and enhanced compatibility with MLLM. Lastly, we designed a multi-stage training strategy that progressively integrates multimodal alignment and multitask fine-tuning, ensuring effective synergy across all modalities. Baichuan-Omni-1.5 leads contemporary models (including GPT4o-mini and MiniCPM-o 2.6) in terms of comprehensive omni-modal capabilities. Notably, it achieves results comparable to leading models such as Qwen2-VL-72B across various multimodal medical benchmarks.

Sparse random mode decomposition (SRMD) is a novel algorithm that constructs a random time-frequency feature space to sparsely approximate spectrograms, effectively separating modes. However, it fails to distinguish adjacent or overlapped frequency components, especially, those with crossover instantaneous frequencies. To address this limitation, an enhanced version, termed three-dimensional SRMD (3D-SRMD), is proposed in this letter. In 3D-SRMD, the random features are lifted from a two-dimensional space to a three-dimensional (3D) space by introducing one extra chirp rate axis. This enhancement effectively disentangles the frequency components overlapped in the low dimension. Additionally, a novel random feature generation strategy is designed to improve the separation accuracy of 3D-SRMD by combining the 3D ridge detection method. Finally, numerical experiments on both simulated and real-world signals demonstrate the effectiveness of our method.

In human-centric scenes, the ability to simultaneously understand visual and auditory information is crucial. While recent omni models can process multiple modalities, they generally lack effectiveness in human-centric scenes due to the absence of large-scale, specialized datasets and non-targeted architectures. In this work, we developed HumanOmni, the industry's first human-centric Omni-multimodal large language model. We constructed a dataset containing over 2.4 million human-centric video clips with detailed captions and more than 14 million instructions, facilitating the understanding of diverse human-centric scenes. HumanOmni includes three specialized branches for understanding different types of scenes. It adaptively fuses features from these branches based on user instructions, significantly enhancing visual understanding in scenes centered around individuals. Moreover, HumanOmni integrates audio features to ensure a comprehensive understanding of environments and individuals. Our experiments validate HumanOmni's advanced capabilities in handling human-centric scenes across a variety of tasks, including emotion recognition, facial expression description, and action understanding. Our model will be open-sourced to facilitate further development and collaboration within both academia and industry.

EEG-based fatigue monitoring can effectively reduce the incidence of related traffic accidents. In the past decade, with the advancement of deep learning, convolutional neural networks (CNN) have been increasingly used for EEG signal processing. However, due to the data's non-Euclidean characteristics, existing CNNs may lose important spatial information from EEG, specifically channel correlation. Thus, we propose the node-holistic graph convolutional network (NHGNet), a model that uses graphic convolution to dynamically learn each channel's features. With exact fit attention optimization, the network captures inter-channel correlations through a trainable adjacency matrix. The interpretability is enhanced by revealing critical areas of brain activity and their interrelations in various mental states. In validations on two public datasets, NHGNet outperforms the SOTAs. Specifically, in the intra-subject, NHGNet improved detection accuracy by at least 2.34% and 3.42%, and in the inter-subjects, it improved by at least 2.09% and 15.06%. Visualization research on the model revealed that the central parietal area plays an important role in detecting fatigue levels, whereas the frontal and temporal lobes are essential for maintaining vigilance.

Large language models (LLMs) are expected to offer structured Markdown responses for the sake of readability in web chatbots (e.g., ChatGPT). Although there are a myriad of metrics to evaluate LLMs, they fail to evaluate the readability from the view of output content structure. To this end, we focus on an overlooked yet important metric -- Markdown Awareness, which directly impacts the readability and structure of the content generated by these language models. In this paper, we introduce MDEval, a comprehensive benchmark to assess Markdown Awareness for LLMs, by constructing a dataset with 20K instances covering 10 subjects in English and Chinese. Unlike traditional model-based evaluations, MDEval provides excellent interpretability by combining model-based generation tasks and statistical methods. Our results demonstrate that MDEval achieves a Spearman correlation of 0.791 and an accuracy of 84.1% with human, outperforming existing methods by a large margin. Extensive experimental results also show that through fine-tuning over our proposed dataset, less performant open-source models are able to achieve comparable performance to GPT-4o in terms of Markdown Awareness. To ensure reproducibility and transparency, MDEval is open sourced at https://github.com/SWUFE-DB-Group/MDEval-Benchmark., Comment: WWW 2025

The first evidence for the decay $B^-\rightarrow D^{**0}\tau^-\overline{\nu_{\tau}}$ is obtained using proton-proton collision data collected by the LHCb experiment, corresponding to an integrated luminosity of 9 fb$^{-1}$ , at centre-of-mass energies of 7, 8 and 13 Tev. Here, the $D^{**0}$ meson represents any of the three excited charm mesons $D_{1}(2420)^{0}$, $D_{2}^{*}(2460)^{0}$, and $D_{1}^{'}(2400)^{0}$. The $B^-\rightarrow D^{**0}\tau^-\overline{\nu_{\tau}}$ signal is measured with a significance of 3.5 $\sigma$, including systematic uncertainties. The combined branching fraction $BR(B^-\rightarrow D^{**0}_{1,2}\tau^-\overline{\nu_{\tau}})\times BR(D^{**0}_{1,2}\rightarrow D^{*+}\pi^-)$, where $D^{**0}_{1,2}$ denotes both $D_{1}(2420)^{0}$ and $D_{2}^{*}(2460)^{0}$ contributions, is measured to be $(0.051\pm0.013(stat)\pm 0.006(syst)\pm 0.009(\rm{ext}) )\%$, where the last uncertainty reflects that of the branching fraction of the normalisation channel $B^-\rightarrow D^{**0}_{1,2}D_s^{(*)-}$. The ratio between the tauonic and muonic semileptonic $B$ decays, with the latter taken from world average values, is also determined and found to be ${\cal R}(D^{**0}_{1,2})=0.13\pm0.03(stat)\pm0.01(syst)\pm0.02\,(\rm{ext})$., Comment: All figures and tables, along with machine-readable versions and any supplementary material and additional information, are available at https://lbfence.cern.ch/alcm/public/analysis/full-details/3300/ (LHCb public pages)

Unmanned Aerial Vehicles (UAVs) are pivotal in applications such as search and rescue and environmental monitoring, excelling in intelligent perception tasks. However, their limited battery capacity hinders long-duration and long-distance missions. Charging UAVs (CUAVs) offers a potential solution by recharging mission UAVs (MUAVs), but existing methods rely on impractical pre-planned routes, failing to enable organic cooperation and limiting mission efficiency. We introduce a novel multi-agent deep reinforcement learning model named \textbf{H}eterogeneous \textbf{G}raph \textbf{A}ttention \textbf{M}ulti-agent Deep Deterministic Policy Gradient (HGAM), designed to dynamically coordinate MUAVs and CUAVs. This approach maximizes data collection, geographical fairness, and energy efficiency by allowing UAVs to adapt their routes in real-time to current task demands and environmental conditions without pre-planning. Our model uses heterogeneous graph attention networks (GATs) to present heterogeneous agents and facilitate efficient information exchange. It operates within an actor-critic framework. Simulation results show that our model significantly improves cooperation among heterogeneous UAVs, outperforming existing methods in several metrics, including data collection rate and charging efficiency.

This paper introduces DeepFlow, a scalable and serverless AI platform designed to efficiently serve large language models (LLMs) at scale in cloud environments. DeepFlow addresses key challenges such as resource allocation, serving efficiency, and cold start latencies through four main design components. First, it uses a simple serverless abstraction called the request-job-task model, which helps manage AI workloads across post-training and model serving tasks. Second, it builds an in-house serving engine FlowServe using a microkernel-inspired design, NPU-centric execution, and SPMD-based parallelism to optimize LLM serving. The system also includes novel scheduling policies tailored for both PD-disaggregated and PD-colocated configurations. With optimizations like pre-warmed pods, DRAM pre-loading, and NPU-fork, DeepFlow can scale up to 64 instances in seconds. DeepFlow has been in production for over a year, operating on a large Ascend NPU cluster and providing industrystandard APIs for fine-tuning, agent serving, and model serving to our customers.

In E-commerce platforms, a full advertising image is composed of a background image and marketing taglines. Automatic ad image design reduces human costs and plays a crucial role. For the convenience of users, a novel automatic framework named Product-Centric Advertising Image Design (PAID) is proposed in this work. PAID takes the product foreground image, required taglines, and target size as input and creates an ad image automatically. PAID consists of four sequential stages: prompt generation, layout generation, background image generation, and graphics rendering. Different expert models are trained to conduct these sub-tasks. A visual language model (VLM) based prompt generation model is leveraged to produce a product-matching background prompt. The layout generation model jointly predicts text and image layout according to the background prompt, product, and taglines to achieve the best harmony. An SDXL-based layout-controlled inpainting model is trained to generate an aesthetic background image. Previous ad image design methods take a background image as input and then predict the layout of taglines, which limits the spatial layout due to fixed image content. Innovatively, our PAID adjusts the stages to produce an unrestricted layout. To complete the PAID framework, we created two high-quality datasets, PITA and PIL. Extensive experimental results show that PAID creates more visually pleasing advertising images than previous methods.

Benchmarks are important tools for tracking the rapid advancements in large language model (LLM) capabilities. However, benchmarks are not keeping pace in difficulty: LLMs now achieve over 90\% accuracy on popular benchmarks like MMLU, limiting informed measurement of state-of-the-art LLM capabilities. In response, we introduce Humanity's Last Exam (HLE), a multi-modal benchmark at the frontier of human knowledge, designed to be the final closed-ended academic benchmark of its kind with broad subject coverage. HLE consists of 3,000 questions across dozens of subjects, including mathematics, humanities, and the natural sciences. HLE is developed globally by subject-matter experts and consists of multiple-choice and short-answer questions suitable for automated grading. Each question has a known solution that is unambiguous and easily verifiable, but cannot be quickly answered via internet retrieval. State-of-the-art LLMs demonstrate low accuracy and calibration on HLE, highlighting a significant gap between current LLM capabilities and the expert human frontier on closed-ended academic questions. To inform research and policymaking upon a clear understanding of model capabilities, we publicly release HLE at https://lastexam.ai., Comment: 25 pages, 6 figures

As the demand for high-resolution image processing in Large Vision-Language Models (LVLMs) grows, sub-image partitioning has become a popular approach for mitigating visual information loss associated with fixed-resolution processing. However, existing partitioning methods uniformly process sub-images, resulting in suboptimal image understanding. In this work, we reveal that the sub-images with higher semantic relevance to the entire image encapsulate richer visual information for preserving the model's visual understanding ability. Therefore, we propose the Global Semantic-guided Weight Allocator (GSWA) module, which dynamically allocates weights to sub-images based on their relative information density, emulating human visual attention mechanisms. This approach enables the model to focus on more informative regions, overcoming the limitations of uniform treatment. We integrate GSWA into the InternVL2-2B framework to create SleighVL, a lightweight yet high-performing model. Extensive experiments demonstrate that SleighVL outperforms models with comparable parameters and remains competitive with larger models. Our work provides a promising direction for more efficient and contextually aware high-resolution image processing in LVLMs, advancing multimodal system development., Comment: 10 pages, 10 figures and tables

Mamba is an efficient sequence model that rivals Transformers and demonstrates significant potential as a foundational architecture for various tasks. Quantization is commonly used in neural networks to reduce model size and computational latency. However, applying quantization to Mamba remains underexplored, and existing quantization methods, which have been effective for CNN and Transformer models, appear inadequate for Mamba models (e.g., Quarot suffers a 21% accuracy drop on Vim-T$^\dagger$ even under W8A8). We have pioneered the exploration of this issue and identified several key challenges. First, significant outliers are present in gate projections, output projections, and matrix multiplications. Second, Mamba's unique parallel scan further amplifies these outliers, leading to uneven and heavy-tailed data distributions. Third, even with the application of the Hadamard transform, the variance across channels in weights and activations still remains inconsistent. To these ends, we propose MambaQuant, a post-training quantization (PTQ) framework consisting of: 1) Karhunen-Loeve Transformation (KLT) enhanced rotation, rendering the rotation matrix adaptable to diverse channel distributions. 2) Smooth-Fused rotation, which equalizes channel variances and can merge additional parameters into model weights. Experiments show that MambaQuant can quantize both weights and activations into 8-bit with less than 1% accuracy loss for Mamba-based vision and language tasks. To the best of our knowledge, MambaQuant is the first comprehensive PTQ design for the Mamba family, paving the way for further advancements in its application.

Using data samples collected by the \mbox{BESIII} detector located at the Beijing Electron Positron Collider, the cross sections of the process $e^+e^-\to f_{1}(1285)\pi^+\pi^-$ are measured at forty-five center-of-mass energies from $3.808$ to $4.951 {\rm GeV}$. An investigation on the cross section line shape is performed, and no significant structure is observed.