arXiv subjects

Xiang Li

Publications and source records attributed to Xiang Li.

At least 559 records · Page 31Linked to original sources

arXiv2025

Enhancing Large-Scale AI Training Efficiency: The C4 Solution for Real-Time Anomaly Detection and Communication Optimization

The emergence of Large Language Models (LLMs) has necessitated the adoption of distributed training techniques, involving the deployment of thousands of GPUs to train a single model. Unfortunately, the efficiency of large-scale distributed training systems is often suboptimal due to the increased likelihood of hardware errors in high-end GPU products and the heightened risk of network traffic collisions. Moreover, any local hardware failure can disrupt training tasks, and the inability to swiftly identify faulty components leads to a significant waste of GPU resources. And, prolonged communication due to traffic collisions can substantially increase GPU waiting times. To address these challenges, we propose a communication-driven solution, namely the C4. The key insights of C4 are twofold. First, the load in distributed training exhibits homogeneous characteristics and is divided into iterations through periodic synchronization, therefore hardware anomalies would incur certain syndrome in collective communication. By leveraging this feature, C4 can rapidly identify the faulty components, swiftly isolate the anomaly, and restart the task, thereby avoiding resource wastage caused by delays in anomaly detection. Second, the predictable communication model of collective communication, involving a limited number of long-lived flows, allows C4 to efficiently execute traffic planning, substantially reducing bandwidth competition among these flows. The C4 has been extensively deployed across real-world production systems in a hyperscale cloud provider, yielding a significant improvement in system efficiency, from 30% to 45%. This enhancement is attributed to a 30% reduction in error-induced overhead and a 15% reduction in communication costs.

cs.DC↗

arXiv2025

DanceTogether! Identity-Preserving Multi-Person Interactive Video Generation

Controllable video generation (CVG) has advanced rapidly, yet current systems falter when more than one actor must move, interact, and exchange positions under noisy control signals. We address this gap with DanceTogether, the first end-to-end diffusion framework that turns a single reference image plus independent pose-mask streams into long, photorealistic videos while strictly preserving every identity. A novel MaskPoseAdapter binds "who" and "how" at every denoising step by fusing robust tracking masks with semantically rich-but noisy-pose heat-maps, eliminating the identity drift and appearance bleeding that plague frame-wise pipelines. To train and evaluate at scale, we introduce (i) PairFS-4K, 26 hours of dual-skater footage with 7,000+ distinct IDs, (ii) HumanRob-300, a one-hour humanoid-robot interaction set for rapid cross-domain transfer, and (iii) TogetherVideoBench, a three-track benchmark centered on the DanceTogEval-100 test suite covering dance, boxing, wrestling, yoga, and figure skating. On TogetherVideoBench, DanceTogether outperforms the prior arts by a significant margin. Moreover, we show that a one-hour fine-tune yields convincing human-robot videos, underscoring broad generalization to embodied-AI and HRI tasks. Extensive ablations confirm that persistent identity-action binding is critical to these gains. Together, our model, datasets, and benchmark lift CVG from single-subject choreography to compositionally controllable, multi-actor interaction, opening new avenues for digital production, simulation, and embodied intelligence. Our video demos and code are available at https://DanceTog.github.io/.

cs.CV↗

arXiv2025

ASMA-Tune: Unlocking LLMs' Assembly Code Comprehension via Structural-Semantic Instruction Tuning

Assembly code analysis and comprehension play critical roles in applications like reverse engineering, yet they face substantial challenges due to low information density and a lack of explicit syntactic structures. While traditional masked language modeling (MLM) approaches do not explicitly focus on natural language interaction, emerging decoder-focused large language models (LLMs) demonstrate partial success in binary analysis yet remain underexplored for holistic comprehension. We present Assembly Augmented Tuning, an end-to-end structural-semantic instruction tuning framework that synergizes encoder architecture with decoder-based LLMs through a projector module, where the assembly encoder extracts hardware-level structural features, the projector bridges representations with the semantic space, and the instruction-tuned LLM preserves natural language capabilities. Experimental results demonstrate three key advantages: (1) State-of-the-art performance in assembly comprehension with +39.7% Recall@1 and +17.8% MRR improvements over GPT-4-Turbo, (2) Consistent enhancements across base models (24.6-107.4% Recall@1 and 15.2-106.3% MRR on Qwen2.5-Coder, Deepseek-Coder and CodeLlama variants), and (3) Superior instruction-following capabilities (41.5%-118% improvements) with controlled code generation degradation (-8.9% to -35% across architectures).

cs.SE↗

arXiv2025

Enhancement of temperature of quantum anomalous Hall effect in two-dimensional germanene/magnetic semiconductor heterostructures

Quantum anomalous Hall effect (QAHE) is significant for future low-power electronics devices, where a main challenge is realizing QAHE at high temperatures. In this work, based on experimentally reported two-dimensional (2D) germanene and magnetic semiconductors Cr$_2$Ge$_2$Te$_6$ and Cr$_2$Si$_2$Te$_6$, and the first principle calculations, germanene/magnetic semiconductor heterostructures are investigated. Topologically nontrivial edge states and quantized anomalous Hall conductance are demonstrated. It is shown that the QAHE temperature can be enhanced to approximately 62 K in germanene/monolayer (ML) Cr$_2$Ge$_2$Te$_6$ with 2.1\% tensile strain, 64 K in germanene/bilayer (BL) Cr$_2$Ge$_2$Te$_6$ with 1.4\% tensile strain, and 50 K in germanene/ML Cr$_2$Si$_2$Te$_6$ with 1.3\% tensile strain. With increasing tensile strain of these heterostructures, the band gap decreases and the Curie temperature rises, and the highest temperature of QAHE is obtained. Since these 2D materials were discovered in recent experiments, our results provide promising materials for achieving high-temperature QAHE.

cond-mat.mes-hall↗

arXiv2025

GMM-Based Comprehensive Feature Extraction and Relative Distance Preservation For Few-Shot Cross-Modal Retrieval

Few-shot cross-modal retrieval focuses on learning cross-modal representations with limited training samples, enabling the model to handle unseen classes during inference. Unlike traditional cross-modal retrieval tasks, which assume that both training and testing data share the same class distribution, few-shot retrieval involves data with sparse representations across modalities. Existing methods often fail to adequately model the multi-peak distribution of few-shot cross-modal data, resulting in two main biases in the latent semantic space: intra-modal bias, where sparse samples fail to capture intra-class diversity, and inter-modal bias, where misalignments between image and text distributions exacerbate the semantic gap. These biases hinder retrieval accuracy. To address these issues, we propose a novel method, GCRDP, for few-shot cross-modal retrieval. This approach effectively captures the complex multi-peak distribution of data using a Gaussian Mixture Model (GMM) and incorporates a multi-positive sample contrastive learning mechanism for comprehensive feature modeling. Additionally, we introduce a new strategy for cross-modal semantic alignment, which constrains the relative distances between image and text feature distributions, thereby improving the accuracy of cross-modal representations. We validate our approach through extensive experiments on four benchmark datasets, demonstrating superior performance over six state-of-the-art methods.

cs.CV↗

arXiv2025

MUBox: A Critical Evaluation Framework of Deep Machine Unlearning

Recent legal frameworks have mandated the right to be forgotten, obligating the removal of specific data upon user requests. Machine Unlearning has emerged as a promising solution by selectively removing learned information from machine learning models. This paper presents MUBox, a comprehensive platform designed to evaluate unlearning methods in deep learning. MUBox integrates 23 advanced unlearning techniques, tested across six practical scenarios with 11 diverse evaluation metrics. It allows researchers and practitioners to (1) assess and compare the effectiveness of different machine unlearning methods across various scenarios; (2) examine the impact of current evaluation metrics on unlearning performance; and (3) conduct detailed comparative studies on machine unlearning in a unified framework. Leveraging MUBox, we systematically evaluate these unlearning methods in deep learning and uncover several key insights: (a) Even state-of-the-art unlearning methods, including those published in top-tier venues and winners of unlearning competitions, demonstrate inconsistent effectiveness across diverse scenarios. Prior research has predominantly focused on simplified settings, such as random forgetting and class-wise unlearning, highlighting the need for broader evaluations across more difficult unlearning tasks. (b) Assessing unlearning performance remains a non-trivial problem, as no single evaluation metric can comprehensively capture the effectiveness, efficiency, and preservation of model utility. Our findings emphasize the necessity of employing multiple metrics to achieve a balanced and holistic assessment of unlearning methods. (c) In the context of depoisoning, our evaluation reveals significant variability in the effectiveness of existing approaches, which is highly dependent on the specific type of poisoning attacks.

cs.LG↗

arXiv2025

GeoGround: A Unified Large Vision-Language Model for Remote Sensing Visual Grounding

Remote sensing (RS) visual grounding aims to use natural language expression to locate specific objects (in the form of the bounding box or segmentation mask) in RS images, enhancing human interaction with intelligent RS interpretation systems. Early research in this area was primarily based on horizontal bounding boxes (HBBs), but as more diverse RS datasets have become available, tasks involving oriented bounding boxes (OBBs) and segmentation masks have emerged. In practical applications, different targets require different grounding types: HBB can localize an object's position, OBB provides its orientation, and mask depicts its shape. However, existing specialized methods are typically tailored to a single type of RS visual grounding task and are hard to generalize across tasks. In contrast, large vision-language models (VLMs) exhibit powerful multi-task learning capabilities but struggle to handle dense prediction tasks like segmentation. This paper proposes GeoGround, a novel framework that unifies support for HBB, OBB, and mask RS visual grounding tasks, allowing flexible output selection. Rather than customizing the architecture of VLM, our work aims to elegantly support pixel-level visual grounding output through the Text-Mask technique. We define prompt-assisted and geometry-guided learning to enhance consistency across different signals. Experimental results show that GeoGround demonstrates strong performance across four RS visual grounding tasks, matching the performance of specialized methods on multiple benchmarks. Code available at https://github.com/zytx121/GeoGround

cs.CV↗

arXiv2025

A Simple Detector with Frame Dynamics is a Strong Tracker

Infrared object tracking plays a crucial role in Anti-Unmanned Aerial Vehicle (Anti-UAV) applications. Existing trackers often depend on cropped template regions and have limited motion modeling capabilities, which pose challenges when dealing with tiny targets. To address this, we propose a simple yet effective infrared tiny-object tracker that enhances tracking performance by integrating global detection and motion-aware learning with temporal priors. Our method is based on object detection and achieves significant improvements through two key innovations. First, we introduce frame dynamics, leveraging frame difference and optical flow to encode both prior target features and motion characteristics at the input level, enabling the model to better distinguish the target from background clutter. Second, we propose a trajectory constraint filtering strategy in the post-processing stage, utilizing spatio-temporal priors to suppress false positives and enhance tracking robustness. Extensive experiments show that our method consistently outperforms existing approaches across multiple metrics in challenging infrared UAV tracking scenarios. Notably, we achieve state-of-the-art performance in the 4th Anti-UAV Challenge, securing 1st place in Track 1 and 2nd place in Track 2.

cs.CV↗

arXiv2025

Robust Model-Based In-Hand Manipulation with Integrated Real-Time Motion-Contact Planning and Tracking

Robotic dexterous in-hand manipulation, where multiple fingers dynamically make and break contact, represents a step toward human-like dexterity in real-world robotic applications. Unlike learning-based approaches that rely on large-scale training or extensive data collection for each specific task, model-based methods offer an efficient alternative. Their online computing nature allows for ready application to new tasks without extensive retraining. However, due to the complexity of physical contacts, existing model-based methods encounter challenges in efficient online planning and handling modeling errors, which limit their practical applications. To advance the effectiveness and robustness of model-based contact-rich in-hand manipulation, this paper proposes a novel integrated framework that mitigates these limitations. The integration involves two key aspects: 1) integrated real-time planning and tracking achieved by a hierarchical structure; and 2) joint optimization of motions and contacts achieved by integrated motion-contact modeling. Specifically, at the high level, finger motion and contact force references are jointly generated using contact-implicit model predictive control. The high-level module facilitates real-time planning and disturbance recovery. At the low level, these integrated references are concurrently tracked using a hand force-motion model and actual tactile feedback. The low-level module compensates for modeling errors and enhances the robustness of manipulation. Extensive experiments demonstrate that our approach outperforms existing model-based methods in terms of accuracy, robustness, and real-time performance. Our method successfully completes five challenging tasks in real-world environments, even under appreciable external disturbances.

cs.RO↗

arXiv2025

High Curie temperature in diluted magnetic semiconductors (B, Mn)X (X = N, P, As, Sb)

Doping nonmagnetic semiconductors with magnetic impurities is a feasible way to obtain diluted magnetic semiconductors (DMSs). It is generally accepted that for the most extensively studied DMS, (Ga, Mn)As, its highest Curie temperature T$_{\text{C}}$ was achieved at 200 K with a Mn concentration of approximately 16\% in experiments. A recent experiment reported record-breaking high electron and hole mobilities in the semiconductor BAs [\href{https://www.science.org/doi/10.1126/science.abn4290}{Science 377, 437 (2022)}]. Since BAs shares the same zinc-blende structure with GaAs, here we predict four DMSs (B, Mn)X (X $=$ N, P, As, Sb) by density functional theory calculations. Using a rescaling method to diminish the overestimation of Curie temperature, our results indicate that a significantly higher T$_{\text{C}}$ in the range of 467 K to 485 K for (B, Mn)As with a Mn concentration of around 15.6\% and even higher T$_{\text{C}}$ values above the room temperature for (B, Mn)P with a Mn concentration exceeding 9.4\%. Furthermore, using the method of Ab initio Scattering and Transport (AMSET) with first-principles material parameters, we have predicted a hole mobility of 48.9 cm$^{\text{2}}$V$^{\text{-1}}$s$^{\text{-1}}$ at 300 K for (B, Mn)As with the hole concentration of about n$_{\text{h}}$ $=$ 4.2 $\times$ 10$^{\text{19}}$ cm$^{\text{-3}}$, which is about two times larger than the hole mobility at 300 K in the calculations for (Ga, Mn)As. The hole mobility of (B, Mn)As can be enhanced faster than that of (Ga, Mn)As when the hole concentration is decreased. Our findings predict the emergence of a new family of DMS, (B, Mn)X, and are expected to stimulate both experimental and theoretical studies of the DMS with possible high T$_{\text{C}}$.

cond-mat.mtrl-sci↗

arXiv2025

Towards Initialization-Agnostic Clustering with Iterative Adaptive Resonance Theory

The clustering performance of Fuzzy Adaptive Resonance Theory (Fuzzy ART) is highly dependent on the preset vigilance parameter, where deviations in its value can lead to significant fluctuations in clustering results, severely limiting its practicality for non-expert users. Existing approaches generally enhance vigilance parameter robustness through adaptive mechanisms such as particle swarm optimization and fuzzy logic rules. However, they often introduce additional hyperparameters or complex frameworks that contradict the original simplicity of the algorithm. To address this, we propose Iterative Refinement Adaptive Resonance Theory (IR-ART), which integrates three key phases into a unified iterative framework: (1) Cluster Stability Detection: A dynamic stability detection module that identifies unstable clusters by analyzing the change of sample size (number of samples in the cluster) in iteration. (2) Unstable Cluster Deletion: An evolutionary pruning module that eliminates low-quality clusters. (3) Vigilance Region Expansion: A vigilance region expansion mechanism that adaptively adjusts similarity thresholds. Independent of the specific execution of clustering, these three phases sequentially focus on analyzing the implicit knowledge within the iterative process, adjusting weights and vigilance parameters, thereby laying a foundation for the next iteration. Experimental evaluation on 15 datasets demonstrates that IR-ART improves tolerance to suboptimal vigilance parameter values while preserving the parameter simplicity of Fuzzy ART. Case studies visually confirm the algorithm's self-optimization capability through iterative refinement, making it particularly suitable for non-expert users in resource-constrained scenarios.

cs.LG↗

arXiv2025

Tamed Euler-Maruyama method for SDEs with non-globally Lipschitz drift and multiplicative noise

Consider the following stochastic differential equation driven by multiplicative noise on $\mathbb{R}^d$ with a superlinearly growing drift coefficient, \begin{align*} \mathrm{d} X_t = b (X_t) \, \mathrm{d} t + σ(X_t) \, \mathrm{d} B_t. \end{align*} It is known that the corresponding explicit Euler schemes may not converge. In this article, we analyze an explicit and easily implementable numerical method for approximating such a stochastic differential equation, i.e. its tamed Euler-Maruyama approximation. Under partial dissipation conditions ensuring the ergodicity, we obtain the uniform-in-time convergence rates of the tamed Euler-Maruyama process under $L^{1}$-Wasserstein distance and total variation distance.

math.PR↗

arXiv2025

Frog Soup: Zero-Shot, In-Context, and Sample-Efficient Frogger Agents

One of the primary aspirations in reinforcement learning research is developing general-purpose agents capable of rapidly adapting to and mastering novel tasks. While RL gaming agents have mastered many Atari games, they remain slow and costly to train for each game. In this work, we demonstrate that latest reasoning LLMs with out-of-domain RL post-training can play a challenging Atari game called Frogger under a zero-shot setting. We then investigate the effect of in-context learning and the amount of reasoning effort on LLM performance. Lastly, we demonstrate a way to bootstrap traditional RL method with LLM demonstrations, which significantly improves their performance and sample efficiency. Our implementation is open sourced at https://github.com/AlienKevin/frogger.

cs.AI↗

arXiv2025

VideoMultiAgents: A Multi-Agent Framework for Video Question Answering

Video Question Answering (VQA) inherently relies on multimodal reasoning, integrating visual, temporal, and linguistic cues to achieve a deeper understanding of video content. However, many existing methods rely on feeding frame-level captions into a single model, making it difficult to adequately capture temporal and interactive contexts. To address this limitation, we introduce VideoMultiAgents, a framework that integrates specialized agents for vision, scene graph analysis, and text processing. It enhances video understanding leveraging complementary multimodal reasoning from independently operating agents. Our approach is also supplemented with a question-guided caption generation, which produces captions that highlight objects, actions, and temporal transitions directly relevant to a given query, thus improving the answer accuracy. Experimental results demonstrate that our method achieves state-of-the-art performance on Intent-QA (79.0%, +6.2% over previous SOTA), EgoSchema subset (75.4%, +3.4%), and NExT-QA (79.6%, +0.4%). The source code is available at https://github.com/PanasonicConnect/VideoMultiAgents.

cs.CV↗

arXiv2025

Ascendra: Dynamic Request Prioritization for Efficient LLM Serving

The rapid advancement of Large Language Models (LLMs) has driven the need for more efficient serving strategies. In this context, efficiency refers to the proportion of requests that meet their Service Level Objectives (SLOs), particularly for Time To First Token (TTFT) and Time Between Tokens (TBT). However, existing systems often prioritize one metric at the cost of the other. We present Ascendra, an LLM serving system designed to meet both TTFT and TBT SLOs simultaneously. The core insight behind Ascendra is that a request's urgency evolves as it approaches its deadline. To leverage this, Ascendra partitions GPU resources into two types of instances: low-priority and high-priority. Low-priority instances maximize throughput by processing requests out of arrival order, but at the risk of request starvation. To address this, Ascendra employs a performance model to predict requests at risk of missing their SLOs and proactively offloads them to high-priority instances. High-priority instances are optimized for low-latency execution and handle urgent requests nearing their deadlines. This partitioned architecture enables Ascendra to effectively balance high throughput and low latency. Extensive evaluation shows that Ascendra improves system throughput by up to 1.7x compared to vLLM and Sarathi-Serve while meeting both TTFT and TBT SLOs.

cs.AI↗

arXiv2025

Seed1.5-Thinking: Advancing Superb Reasoning Models with Reinforcement Learning

We introduce Seed1.5-Thinking, capable of reasoning through thinking before responding, resulting in improved performance on a wide range of benchmarks. Seed1.5-Thinking achieves 86.7 on AIME 2024, 55.0 on Codeforces and 77.3 on GPQA, demonstrating excellent reasoning abilities in STEM and coding. Beyond reasoning tasks, the method demonstrates notable generalization across diverse domains. For instance, it surpasses DeepSeek R1 by 8% in win rate on non-reasoning tasks, indicating its broader applicability. Compared to other state-of-the-art reasoning models, Seed1.5-Thinking is a Mixture-of-Experts (MoE) model with a relatively small size, featuring 20B activated and 200B total parameters. As part of our effort to assess generalized reasoning, we develop two internal benchmarks, BeyondAIME and Codeforces, both of which will be publicly released to support future research. Model trial link: https://www.volcengine.com/experience/ark.

cs.CL↗

arXiv2025

3DCoMPaT$^{++}$: An improved Large-scale 3D Vision Dataset for Compositional Recognition

In this work, we present 3DCoMPaT$^{++}$, a multimodal 2D/3D dataset with 160 million rendered views of more than 10 million stylized 3D shapes carefully annotated at the part-instance level, alongside matching RGB point clouds, 3D textured meshes, depth maps, and segmentation masks. 3DCoMPaT$^{++}$ covers 41 shape categories, 275 fine-grained part categories, and 293 fine-grained material classes that can be compositionally applied to parts of 3D objects. We render a subset of one million stylized shapes from four equally spaced views as well as four randomized views, leading to a total of 160 million renderings. Parts are segmented at the instance level, with coarse-grained and fine-grained semantic levels. We introduce a new task, called Grounded CoMPaT Recognition (GCR), to collectively recognize and ground compositions of materials on parts of 3D objects. Additionally, we report the outcomes of a data challenge organized at CVPR2023, showcasing the winning method's utilization of a modified PointNet$^{++}$ model trained on 6D inputs, and exploring alternative techniques for GCR enhancement. We hope our work will help ease future research on compositional 3D Vision.

cs.CV↗

arXiv2025

Combating the Bucket Effect:Multi-Knowledge Alignment for Medication Recommendation

Medication recommendation is crucial in healthcare, offering effective treatments based on patient's electronic health records (EHR). Previous studies show that integrating more medication-related knowledge improves medication representation accuracy. However, not all medications encompass multiple types of knowledge data simultaneously. For instance, some medications provide only textual descriptions without structured data. This imbalance in data availability limits the performance of existing models, a challenge we term the "bucket effect" in medication recommendation. Our data analysis uncovers the severity of the "bucket effect" in medication recommendation. To fill this gap, we introduce a cross-modal medication encoder capable of seamlessly aligning data from different modalities and propose a medication recommendation framework to integrate Multiple types of Knowledge, named MKMed. Specifically, we first pre-train a cross-modal encoder with contrastive learning on five knowledge modalities, aligning them into a unified space. Then, we combine the multi-knowledge medication representations with patient records for recommendations. Extensive experiments on the MIMIC-III and MIMIC-IV datasets demonstrate that MKMed mitigates the "bucket effect" in data, and significantly outperforms state-of-the-art baselines in recommendation accuracy and safety.

cs.AI↗