arXiv subjects

Yi Huang

Publications and source records attributed to Yi Huang.

At least 91 records · Page 5Linked to original sources

arXiv2025

DIVE: Taming DINO for Subject-Driven Video Editing

Building on the success of diffusion models in image generation and editing, video editing has recently gained substantial attention. However, maintaining temporal consistency and motion alignment still remains challenging. To address these issues, this paper proposes DINO-guided Video Editing (DIVE), a framework designed to facilitate subject-driven editing in source videos conditioned on either target text prompts or reference images with specific identities. The core of DIVE lies in leveraging the powerful semantic features extracted from a pretrained DINOv2 model as implicit correspondences to guide the editing process. Specifically, to ensure temporal motion consistency, DIVE employs DINO features to align with the motion trajectory of the source video. For precise subject editing, DIVE incorporates the DINO features of reference images into a pretrained text-to-image model to learn Low-Rank Adaptations (LoRAs), effectively registering the target subject's identity. Extensive experiments on diverse real-world videos demonstrate that our framework can achieve high-quality editing results with robust motion consistency, highlighting the potential of DINO to contribute to video editing. Project page: https://dino-video-editing.github.io

cs.CV↗

arXiv2025

Knowledge Augmented Finetuning Matters in both RAG and Agent Based Dialog Systems

Large language models (LLMs) have recently been applied to dialog systems. Despite making progress, LLMs are prone to errors in knowledge-intensive scenarios. Recently, approaches based on retrieval augmented generation (RAG) and agent have emerged to improve the factual accuracy by enhancing the LLMs with knowledge retrieved from external knowledge bases (KBs). This is mostly implemented by prompting the LLMs with instructions, examples and the retrieved knowledge. However, LLMs may have difficulty using the retrieved knowledge effectively for response generation, because they are not well trained to do such generation for specific domains. To mitigate this problem, we propose to finetune the LLMs in the RAG-based and agent-based systems with domain-specific data, together with domain-specific external knowledge, which is called knowledge augmented finetuning (KAFT). We base our study on the MobileCS2 dataset, a real-life customer service dialog dataset that features intensive knowledge interactions, to systematically compare the prompting and KAFT techniques in the RAG-based and agent-based systems. Experiment results show that KAFT substantially surpasses prompting in both RAG and agent systems, particularly in terms of factual accuracy. To the best of our knowledge, this paper represents the first solid empirical work to investigate the KAFT idea.

cs.CL↗

arXiv2025

From Superficial to Deep: Integrating External Knowledge for Follow-up Question Generation Using Knowledge Graph and LLM

In a conversational system, dynamically generating follow-up questions based on context can help users explore information and provide a better user experience. Humans are usually able to ask questions that involve some general life knowledge and demonstrate higher order cognitive skills. However, the questions generated by existing methods are often limited to shallow contextual questions that are uninspiring and have a large gap to the human level. In this paper, we propose a three-stage external knowledge-enhanced follow-up question generation method, which generates questions by identifying contextual topics, constructing a knowledge graph (KG) online, and finally combining these with a large language model to generate the final question. The model generates information-rich and exploratory follow-up questions by introducing external common sense knowledge and performing a knowledge fusion operation. Experiments show that compared to baseline models, our method generates questions that are more informative and closer to human questioning levels while maintaining contextual relevance.

cs.AI↗

arXiv2025

OutDreamer: Video Outpainting with a Diffusion Transformer

Video outpainting is a challenging task that generates new video content by extending beyond the boundaries of an original input video, requiring both temporal and spatial consistency. Many state-of-the-art methods utilize latent diffusion models with U-Net backbones but still struggle to achieve high quality and adaptability in generated content. Diffusion transformers (DiTs) have emerged as a promising alternative because of their superior performance. We introduce OutDreamer, a DiT-based video outpainting framework comprising two main components: an efficient video control branch and a conditional outpainting branch. The efficient video control branch effectively extracts masked video information, while the conditional outpainting branch generates missing content based on these extracted conditions. Additionally, we propose a mask-driven self-attention layer that dynamically integrates the given mask information, further enhancing the model's adaptability to outpainting tasks. Furthermore, we introduce a latent alignment loss to maintain overall consistency both within and between frames. For long video outpainting, we employ a cross-video-clip refiner to iteratively generate missing content, ensuring temporal consistency across video clips. Extensive evaluations demonstrate that our zero-shot OutDreamer outperforms state-of-the-art zero-shot methods on widely recognized benchmarks.

cs.CV↗

arXiv2025

VLMInferSlow: Evaluating the Efficiency Robustness of Large Vision-Language Models as a Service

Vision-Language Models (VLMs) have demonstrated great potential in real-world applications. While existing research primarily focuses on improving their accuracy, the efficiency remains underexplored. Given the real-time demands of many applications and the high inference overhead of VLMs, efficiency robustness is a critical issue. However, previous studies evaluate efficiency robustness under unrealistic assumptions, requiring access to the model architecture and parameters -- an impractical scenario in ML-as-a-service settings, where VLMs are deployed via inference APIs. To address this gap, we propose VLMInferSlow, a novel approach for evaluating VLM efficiency robustness in a realistic black-box setting. VLMInferSlow incorporates fine-grained efficiency modeling tailored to VLM inference and leverages zero-order optimization to search for adversarial examples. Experimental results show that VLMInferSlow generates adversarial images with imperceptible perturbations, increasing the computational cost by up to 128.47%. We hope this research raises the community's awareness about the efficiency robustness of VLMs.

cs.CV↗

arXiv2025

Spatiotemporal evolution of PM2.5 diffusion in Cheng-Yu urban agglomeration in response to COVID-19 lockdown using complex network

As the decrease in human activities resulting from the COVID-19 control measures had a significant impact on air quality, the epidemic provided an opportunity to investigate the extent to which air pollution is influenced by human activities and review existing measures. However, the corresponding diffusion pattern on a city scale is seldom mentioned at present stage, therefore, we chose the Cheng-Yu urban agglomeration, which is the largest city cluster in Southwest China, as our study area during the COVID-19 period, and attempted to investigate the process of PM2.5 diffusion using a complex network method. The results displayed that there was an evident external spillover effect of PM2.5 across all regions, and the PM2.5 spillovers were concentrated in several cities in the Cheng-Yu urban agglomeration during the lockdown period, whereas they are more dispersed during the recovery period. The overall decline in the impact of PM2.5 pollution source areas on receptor areas from a normal year to the pandemic year, and the intensity of PM2.5 spillover decreases gradually as the distance from the center increases. The implementation of the lockdown measures had an impact on both the input and output patterns of PM2.5 pollution in the region, the input pattern of PM2.5 pollution exhibited higher vulnerability, while the output pattern showed higher resilience. Additionally, the spillover relationship of PM2.5 pollution varies between different blocks, with relatively simple spillover relationships observed during the lockdown period and more complex dynamics during the recovery period. These findings have highlighted the importance of joint controls in combating regional air pollution.

math.NA↗

arXiv2025

MiMo: Unlocking the Reasoning Potential of Language Model -- From Pretraining to Posttraining

We present MiMo-7B, a large language model born for reasoning tasks, with optimization across both pre-training and post-training stages. During pre-training, we enhance the data preprocessing pipeline and employ a three-stage data mixing strategy to strengthen the base model's reasoning potential. MiMo-7B-Base is pre-trained on 25 trillion tokens, with additional Multi-Token Prediction objective for enhanced performance and accelerated inference speed. During post-training, we curate a dataset of 130K verifiable mathematics and programming problems for reinforcement learning, integrating a test-difficulty-driven code-reward scheme to alleviate sparse-reward issues and employing strategic data resampling to stabilize training. Extensive evaluations show that MiMo-7B-Base possesses exceptional reasoning potential, outperforming even much larger 32B models. The final RL-tuned model, MiMo-7B-RL, achieves superior performance on mathematics, code and general reasoning tasks, surpassing the performance of OpenAI o1-mini. The model checkpoints are available at https://github.com/xiaomimimo/MiMo.

cs.CL↗

arXiv2025

MiMo-VL Technical Report

We open-source MiMo-VL-7B-SFT and MiMo-VL-7B-RL, two powerful vision-language models delivering state-of-the-art performance in both general visual understanding and multimodal reasoning. MiMo-VL-7B-RL outperforms Qwen2.5-VL-7B on 35 out of 40 evaluated tasks, and scores 59.4 on OlympiadBench, surpassing models with up to 78B parameters. For GUI grounding applications, it sets a new standard with 56.1 on OSWorld-G, even outperforming specialized models such as UI-TARS. Our training combines four-stage pre-training (2.4 trillion tokens) with Mixed On-policy Reinforcement Learning (MORL) integrating diverse reward signals. We identify the importance of incorporating high-quality reasoning data with long Chain-of-Thought into pre-training stages, and the benefits of mixed RL despite challenges in simultaneous multi-domain optimization. We also contribute a comprehensive evaluation suite covering 50+ tasks to promote reproducibility and advance the field. The model checkpoints and full evaluation suite are available at https://github.com/XiaomiMiMo/MiMo-VL.

cs.CL↗

arXiv2025

Entriever: Energy-based Retriever for Knowledge-Grounded Dialog Systems

A retriever, which retrieves relevant knowledge pieces from a knowledge base given a context, is an important component in many natural language processing (NLP) tasks. Retrievers have been introduced in knowledge-grounded dialog systems to improve knowledge acquisition. In knowledge-grounded dialog systems, when conditioning on a given context, there may be multiple relevant and correlated knowledge pieces. However, knowledge pieces are usually assumed to be conditionally independent in current retriever models. To address this issue, we propose Entriever, an energy-based retriever. Entriever directly models the candidate retrieval results as a whole instead of modeling the knowledge pieces separately, with the relevance score defined by an energy function. We explore various architectures of energy functions and different training methods for Entriever, and show that Entriever substantially outperforms the strong cross-encoder baseline in knowledge retrieval tasks. Furthermore, we show that in semi-supervised training of knowledge-grounded dialog systems, Entriever enables effective scoring of retrieved knowledge pieces and significantly improves end-to-end performance of dialog systems.

cs.CL↗

arXiv2025

Integer Quantum Hall Effect: Disorder, temperature, floating, and plateau width

We theoretically consider disorder and temperature effects on the integer quantum Hall effect (IQHE) using a variety of distinct and complementary analytical and numerical techniques. In particular, we address simple, physical, and experimentally relevant questions: How does disorder and/or temperature affect the IQHE plateau width? Does the plateau width increase or decrease with disorder and/or temperature? What happens to the peak in the longitudinal conductance with increasing disorder/temperature? Does the longitudinal conductance obey any universal scaling property? Is there "floating" with increasing disorder and/or decreasing magnetic field? Can disorder destroy the IQHE? Is there an IQHE to localization transition? What is the Landau level dependence of the plateau width? Our detailed theory provides answers to these and other related experimentally relevant questions. We discuss our results in the context of existing experimental results and suggest future experiments arising from our work. A key finding is that disorder and temperature are intrinsically connected in affecting IQHE, and there is an intricate interplay between them leading to nonmonotonicity in how the IQHE plateau width behaves as a function of increasing disorder. Both must be considered on an equal footing in understanding IQHE experiments.

cond-mat.mes-hall↗

arXiv2025

EFFACT: A Highly Efficient Full-Stack FHE Acceleration Platform

Fully Homomorphic Encryption (FHE) is a set of powerful cryptographic schemes that allows computation to be performed directly on encrypted data with an unlimited depth. Despite FHE's promising in privacy-preserving computing, yet in most FHE schemes, ciphertext generally blows up thousands of times compared to the original message, and the massive amount of data load from off-chip memory for bootstrapping and privacy-preserving machine learning applications (such as HELR, ResNet-20), both degrade the performance of FHE-based computation. Several hardware designs have been proposed to address this issue, however, most of them require enormous resources and power. An acceleration platform with easy programmability, high efficiency, and low overhead is a prerequisite for practical application. This paper proposes EFFACT, a highly efficient full-stack FHE acceleration platform with a compiler that provides comprehensive optimizations and vector-friendly hardware. We start by examining the computational overhead across different real-world benchmarks to highlight the potential benefits of reallocating computing resources for efficiency enhancement. Then we make a design space exploration to find an optimal SRAM size with high utilization and low cost. On the other hand, EFFACT features a novel optimization named streaming memory access which is proposed to enable high throughput with limited SRAMs. Regarding the software-side optimization, we also propose a circuit-level function unit reuse scheme, to substantially reduce the computing resources without performance degradation. Moreover, we design novel NTT and automorphism units that are suitable for a cost-sensitive and highly efficient architecture, leading to low area. For generality, EFFACT is also equipped with an ISA and a compiler backend that can support several FHE schemes like CKKS, BGV, and BFV.

cs.CR↗

arXiv2025

Chemist-X: Large Language Model-empowered Agent for Reaction Condition Recommendation in Chemical Synthesis

Recent AI research plots a promising future of automatic chemical reactions within the chemistry society. This study proposes Chemist-X, a comprehensive AI agent that automates the reaction condition optimization (RCO) task in chemical synthesis with retrieval-augmented generation (RAG) technology and AI-controlled wet-lab experiment executions. To begin with, as an emulation on how chemical experts solve the RCO task, Chemist-X utilizes a novel RAG scheme to interrogate available molecular and literature databases to narrow the searching space for later processing. The agent then leverages a computer-aided design (CAD) tool we have developed through a large language model (LLM) supervised programming interface. With updated chemical knowledge obtained via RAG, as well as the ability in using CAD tools, our agent significantly outperforms conventional RCO AIs confined to the fixed knowledge within its training data. Finally, Chemist-X interacts with the physical world through an automated robotic system, which can validate the suggested chemical reaction condition without human interventions. The control of the robotic system was achieved with a novel algorithm we have developed for the equipment, which relies on LLMs for reliable script generation. Results of our automatic wet-lab experiments, achieved by fully LLM-supervised end-to-end operation with no human in the lope, prove Chemist-X's ability in self-driving laboratories.

cs.IR↗

arXiv2025

The Tenth NTIRE 2025 Efficient Super-Resolution Challenge Report

This paper presents a comprehensive review of the NTIRE 2025 Challenge on Single-Image Efficient Super-Resolution (ESR). The challenge aimed to advance the development of deep models that optimize key computational metrics, i.e., runtime, parameters, and FLOPs, while achieving a PSNR of at least 26.90 dB on the $\operatorname{DIV2K\_LSDIR\_valid}$ dataset and 26.99 dB on the $\operatorname{DIV2K\_LSDIR\_test}$ dataset. A robust participation saw \textbf{244} registered entrants, with \textbf{43} teams submitting valid entries. This report meticulously analyzes these methods and results, emphasizing groundbreaking advancements in state-of-the-art single-image ESR techniques. The analysis highlights innovative approaches and establishes benchmarks for future research in the field.

cs.CV↗

arXiv2025

Streaming Democratized: Ease Across the Latency Spectrum with Delayed View Semantics and Snowflake Dynamic Tables

Streaming data pipelines remain challenging and expensive to build and maintain, despite significant advancements in stronger consistency, event time semantics, and SQL support over the last decade. Persistent obstacles continue to hinder usability, such as the need for manual incrementalization, semantic discrepancies across SQL implementations, and the lack of enterprise-grade operational features. While the rise of incremental view maintenance (IVM) as a way to integrate streaming with databases has been a huge step forward, transaction isolation in the presence of IVM remains underspecified, leaving the maintenance of application-level invariants as a painful exercise for the user. Meanwhile, most streaming systems optimize for latencies of 100 ms to 3 sec, whereas many practical use cases are well-served by latencies ranging from seconds to tens of minutes. We present delayed view semantics (DVS), a conceptual foundation that bridges the semantic gap between streaming and databases, and introduce Dynamic Tables, Snowflake's declarative streaming transformation primitive designed to democratize analytical stream processing. DVS formalizes the intuition that stream processing is primarily a technique to eagerly compute derived results asynchronously, while also addressing the need to reason about the resulting system end to end. Dynamic Tables then offer two key advantages: ease of use through DVS, enterprise-grade features, and simplicity; as well as scalable cost efficiency via IVM with an architecture designed for diverse latency requirements. We first develop extensions to transaction isolation that permit the preservation of invariants in streaming applications. We then detail the implementation challenges of Dynamic Tables and our experience operating it at scale. Finally, we share insights into user adoption and discuss our vision for the future of stream processing.

cs.DB↗

arXiv2025

HarmonySeg: Tubular Structure Segmentation with Deep-Shallow Feature Fusion and Growth-Suppression Balanced Loss

Accurate segmentation of tubular structures in medical images, such as vessels and airway trees, is crucial for computer-aided diagnosis, radiotherapy, and surgical planning. However, significant challenges exist in algorithm design when faced with diverse sizes, complex topologies, and (often) incomplete data annotation of these structures. We address these difficulties by proposing a new tubular structure segmentation framework named HarmonySeg. First, we design a deep-to-shallow decoder network featuring flexible convolution blocks with varying receptive fields, which enables the model to effectively adapt to tubular structures of different scales. Second, to highlight potential anatomical regions and improve the recall of small tubular structures, we incorporate vesselness maps as auxiliary information. These maps are aligned with image features through a shallow-and-deep fusion module, which simultaneously eliminates unreasonable candidates to maintain high precision. Finally, we introduce a topology-preserving loss function that leverages contextual and shape priors to balance the growth and suppression of tubular structures, which also allows the model to handle low-quality and incomplete annotations. Extensive quantitative experiments are conducted on four public datasets. The results show that our model can accurately segment 2D and 3D tubular structures and outperform existing state-of-the-art methods. External validation on a private dataset also demonstrates good generalizability.

eess.IV↗

arXiv2025

Palette of Language Models: A Solver for Controlled Text Generation

Recent advancements in large language models have revolutionized text generation with their remarkable capabilities. These models can produce controlled texts that closely adhere to specific requirements when prompted appropriately. However, designing an optimal prompt to control multiple attributes simultaneously can be challenging. A common approach is to linearly combine single-attribute models, but this strategy often overlooks attribute overlaps and can lead to conflicts. Therefore, we propose a novel combination strategy inspired by the Law of Total Probability and Conditional Mutual Information Minimization on generative language models. This method has been adapted for single-attribute control scenario and is termed the Palette of Language Models due to its theoretical linkage between attribute strength and generation style, akin to blending colors on an artist's palette. Moreover, positive correlation and attribute enhancement are advanced as theoretical properties to guide a rational combination strategy design. We conduct experiments on both single control and multiple control settings, and achieve surpassing results.

cs.CL↗

arXiv2025

Diffusion Model-Based Image Editing: A Survey

Denoising diffusion models have emerged as a powerful tool for various image generation and editing tasks, facilitating the synthesis of visual content in an unconditional or input-conditional manner. The core idea behind them is learning to reverse the process of gradually adding noise to images, allowing them to generate high-quality samples from a complex distribution. In this survey, we provide an exhaustive overview of existing methods using diffusion models for image editing, covering both theoretical and practical aspects in the field. We delve into a thorough analysis and categorization of these works from multiple perspectives, including learning strategies, user-input conditions, and the array of specific editing tasks that can be accomplished. In addition, we pay special attention to image inpainting and outpainting, and explore both earlier traditional context-driven and current multimodal conditional methods, offering a comprehensive analysis of their methodologies. To further evaluate the performance of text-guided image editing algorithms, we propose a systematic benchmark, EditEval, featuring an innovative metric, LMM Score. Finally, we address current limitations and envision some potential directions for future research. The accompanying repository is released at https://github.com/SiatMMLab/Awesome-Diffusion-Model-Based-Image-Editing-Methods.

cs.CV↗

arXiv2025

Effectiveness of denoising diffusion probabilistic models for fast and high-fidelity whole-event simulation in high-energy heavy-ion experiments

Artificial intelligence (AI) generative models, such as generative adversarial networks (GANs), variational auto-encoders, and normalizing flows, have been widely used and studied as efficient alternatives for traditional scientific simulations. However, they have several drawbacks, including training instability and inability to cover the entire data distribution, especially for regions where data are rare. This is particularly challenging for whole-event, full-detector simulations in high-energy heavy-ion experiments, such as sPHENIX at the Relativistic Heavy Ion Collider and Large Hadron Collider experiments, where thousands of particles are produced per event and interact with the detector. This work investigates the effectiveness of Denoising Diffusion Probabilistic Models (DDPMs) as an AI-based generative surrogate model for the sPHENIX experiment that includes the heavy-ion event generation and response of the entire calorimeter stack. DDPM performance in sPHENIX simulation data is compared with a popular rival, GANs. Results show that both DDPMs and GANs can reproduce the data distribution where the examples are abundant (low-to-medium calorimeter energies). Nonetheless, DDPMs significantly outperform GANs, especially in high-energy regions where data are rare. Additionally, DDPMs exhibit superior stability compared to GANs. The results are consistent between both central and peripheral centrality heavy-ion collision events. Moreover, DDPMs offer a substantial speedup of approximately a factor of 100 compared to the traditional Geant4 simulation method.

physics.data-an↗