Jiahui Chen

Xinyu Zhang, Jiahui Chen, Junkun Yuan et al. · tencent-ai

Masked image modeling (MIM) learns visual representation by masking and reconstructing image patches. Applying the reconstruction supervision on the CLIP representation has been proven effective for MIM. However, it is still under-explored how CLIP supervision in MIM influences performance. To investigate strategies for refining the CLIP-targeted MIM, we study two critical elements in MIM, i.e., the supervision position and the mask ratio, and reveal two interesting perspectives, relying on our developed simple pipeline, context autodecoder with CLIP target (CAE v2). Firstly, we observe that the supervision on visible patches achieves remarkable performance, even better than that on masked patches, where the latter is the standard format in the existing MIM methods. Secondly, the optimal mask ratio positively correlates to the model size. That is to say, the smaller the model, the lower the mask ratio needs to be. Driven by these two discoveries, our simple and concise approach CAE v2 achieves superior performance on a series of downstream tasks. For example, a vanilla ViT-Large model achieves 81.7% and 86.7% top-1 accuracy on linear probing and fine-tuning on ImageNet-1K, and 55.9% mIoU on semantic segmentation on ADE20K with the pre-training for 300 epochs. We hope our findings can be helpful guidelines for the pre-training in the MIM area, especially for the small-scale models.

Xuan Wei, Jiahui Chen, Kaiheng Li et al.

Audio-driven human motion video generation aims to synthesize realistic and temporally coherent human animations from a single static image, with applications in talking-head synthesis, co-speech gesture generation, and dynamic presentations. Moving beyond conventional keypoint-based methods that often struggle to capture subtle motion dynamics, We propose a novel implicit-motion framework for generating realistic and temporally coherent human motion videos from a single static image and audio. Our approach uses a two-stage pipeline that decouples motion prediction from rendering. The first stage integrates appearance priors and hierarchical depth cues into a region-aware attention mechanism to model latent motion features. The second stage employs a Mamba-enhanced diffusion model to directly predict these features from audio and the source image, enabling unsupervised learning of fine-grained motion patterns. This decoupled architecture enhances flexibility and efficiency. Trained on a new 380-hour high-quality dataset, our method outperforms prior work across multiple public benchmarks and our collected data in accuracy, naturalness, and temporal coherence, setting a new state-of-the-art.

Xiaokang Chen, Jiahui Chen, Yan Liu et al.

While various knowledge distillation (KD) methods in CNN-based detectors show their effectiveness in improving small students, the baselines and recipes for DETR-based detectors are yet to be built. In this paper, we focus on the transformer decoder of DETR-based detectors and explore KD methods for them. The outputs of the transformer decoder lie in random order, which gives no direct correspondence between the predictions of the teacher and the student, thus posing a challenge for knowledge distillation. To this end, we propose MixMatcher to align the decoder outputs of DETR-based teachers and students, which mixes two teacher-student matching strategies, i.e., Adaptive Matching and Fixed Matching. Specifically, Adaptive Matching applies bipartite matching to adaptively match the outputs of the teacher and the student in each decoder layer, while Fixed Matching fixes the correspondence between the outputs of the teacher and the student with the same object queries, with the teacher's fixed object queries fed to the decoder of the student as an auxiliary group. Based on MixMatcher, we build \textbf{D}ecoder \textbf{D}istillation for \textbf{DE}tection \textbf{TR}ansformer (D$^3$ETR), which distills knowledge in decoder predictions and attention maps from the teachers to students. D$^3$ETR shows superior performance on various DETR-based detectors with different backbones. For example, D$^3$ETR improves Conditional DETR-R50-C5 by $\textbf{7.8}/\textbf{2.4}$ mAP under $12/50$ epochs training settings with Conditional DETR-R101-C5 as the teacher.

Qiang Chen, Jian Wang, Chuchu Han et al.

We present a strong object detector with encoder-decoder pretraining and finetuning. Our method, called Group DETR v2, is built upon a vision transformer encoder ViT-Huge~\cite{dosovitskiy2020image}, a DETR variant DINO~\cite{zhang2022dino}, and an efficient DETR training method Group DETR~\cite{chen2022group}. The training process consists of self-supervised pretraining and finetuning a ViT-Huge encoder on ImageNet-1K, pretraining the detector on Object365, and finally finetuning it on COCO. Group DETR v2 achieves $\textbf{64.5}$ mAP on COCO test-dev, and establishes a new SoTA on the COCO leaderboard https://paperswithcode.com/sota/object-detection-on-coco

Jiahui Chen, Xu Guo, Wensheng Gan et al.

On-shelf utility mining (OSUM) is an emerging research direction in data mining. It aims to discover itemsets that have high relative utility in their selling time period. Compared with traditional utility mining, OSUM can find more practical and meaningful patterns in real-life applications. However, there is a major drawback to traditional OSUM. For normal users, it is hard to define a minimum threshold minutil for mining the right amount of on-shelf high utility itemsets. On one hand, if the threshold is set too high, the number of patterns would not be enough. On the other hand, if the threshold is set too low, too many patterns will be discovered and cause an unnecessary waste of time and memory consumption. To address this issue, the user usually directly specifies a parameter k, where only the top-k high relative utility itemsets would be considered. Therefore, in this paper, we propose a generic algorithm named TOIT for mining Top-k On-shelf hIgh-utility paTterns to solve this problem. TOIT applies a novel strategy to raise the minutil based on the on-shelf datasets. Besides, two novel upper-bound strategies named subtree utility and local utility are applied to prune the search space. By adopting the strategies mentioned above, the TOIT algorithm can narrow the search space as early as possible, improve the mining efficiency, and reduce the memory consumption, so it can obtain better performance than other algorithms. A series of experiments have been conducted on real datasets with different styles to compare the effects with the state-of-the-art KOSHU algorithm. The experimental results showed that TOIT outperforms KOSHU in both running time and memory consumption.

Jiahui Chen, Yixin Xu, Shicheng Wan et al.

As an important data mining technology, high utility itemset mining (HUIM) is used to find out interesting but hidden information (e.g., profit and risk). HUIM has been widely applied in many application scenarios, such as market analysis, medical detection, and web click stream analysis. However, most previous HUIM approaches often ignore the relationship between items in an itemset. Therefore, many irrelevant combinations (e.g., \{gold, apple\} and \{notebook, book\}) are discovered in HUIM. To address this limitation, many algorithms have been proposed to mine correlated high utility itemsets (CoHUIs). In this paper, we propose a novel algorithm called the Itemset Utility Maximization with Correlation Measure (CoIUM), which considers both a strong correlation and the profitable values of the items. Besides, the novel algorithm adopts a database projection mechanism to reduce the cost of database scanning. Moreover, two upper bounds and four pruning strategies are utilized to effectively prune the search space. And a concise array-based structure named utility-bin is used to calculate and store the adopted upper bounds in linear time and space. Finally, extensive experimental results on dense and sparse datasets demonstrate that CoIUM significantly outperforms the state-of-the-art algorithms in terms of runtime and memory consumption.

Animesh Sinha, Bo Sun, Anmol Kalia et al.

We introduce Style Tailoring, a recipe to finetune Latent Diffusion Models (LDMs) in a distinct domain with high visual quality, prompt alignment and scene diversity. We choose sticker image generation as the target domain, as the images significantly differ from photorealistic samples typically generated by large-scale LDMs. We start with a competent text-to-image model, like Emu, and show that relying on prompt engineering with a photorealistic model to generate stickers leads to poor prompt alignment and scene diversity. To overcome these drawbacks, we first finetune Emu on millions of sticker-like images collected using weak supervision to elicit diversity. Next, we curate human-in-the-loop (HITL) Alignment and Style datasets from model generations, and finetune to improve prompt alignment and style alignment respectively. Sequential finetuning on these datasets poses a tradeoff between better style alignment and prompt alignment gains. To address this tradeoff, we propose a novel fine-tuning method called Style Tailoring, which jointly fits the content and style distribution and achieves best tradeoff. Evaluation results show our method improves visual quality by 14%, prompt alignment by 16.2% and scene diversity by 15.3%, compared to prompt engineering the base Emu model for stickers generation.

Shicheng Wan, Zhenqiang Ye, Wensheng Gan et al.

High utility itemset mining approaches discover hidden patterns from large amounts of temporal data. However, an inescapable problem of high utility itemset mining is that its discovered results hide the quantities of patterns, which causes poor interpretability. The results only reflect the shopping trends of customers, which cannot help decision makers quantify collected information. In linguistic terms, computers use mathematical or programming languages that are precisely formalized, but the language used by humans is always ambiguous. In this paper, we propose a novel one-phase temporal fuzzy utility itemset mining approach called TFUM. It revises temporal fuzzy-lists to maintain less but major information about potential high temporal fuzzy utility itemsets in memory, and then discovers a complete set of real interesting patterns in a short time. In particular, the remaining measure is the first adopted in the temporal fuzzy utility itemset mining domain in this paper. The remaining maximal temporal fuzzy utility is a tighter and stronger upper bound than that of previous studies adopted. Hence, it plays an important role in pruning the search space in TFUM. Finally, we also evaluate the efficiency and effectiveness of TFUM on various datasets. Extensive experimental results indicate that TFUM outperforms the state-of-the-art algorithms in terms of runtime cost, memory usage, and scalability. In addition, experiments prove that the remaining measure can significantly prune unnecessary candidates during mining.

Qi Lin, Wensheng Gan, Yongdong Wu et al.

Nowadays, the environments of smart systems for Industry 4.0 and Internet of Things (IoT) are experiencing fast industrial upgrading. Big data technologies such as design making, event detection, and classification are developed to help manufacturing organizations to achieve smart systems. By applying data analysis, the potential values of rich data can be maximized and thus help manufacturing organizations to finish another round of upgrading. In this paper, we propose two new algorithms with respect to big data analysis, namely UFC$_{gen}$ and UFC$_{fast}$. Both algorithms are designed to collect three types of patterns to help people determine the market positions for different product combinations. We compare these algorithms on various types of datasets, both real and synthetic. The experimental results show that both algorithms can successfully achieve pattern classification by utilizing three different types of interesting patterns from all candidate patterns based on user-specified thresholds of utility and frequency. Furthermore, the list-based UFC$_{fast}$ algorithm outperforms the level-wise-based UFC$_{gen}$ algorithm in terms of both execution time and memory consumption.

Jiahui Chen, Yi Zhao, Qi Li et al.

Deep learning (DL) methods have been widely applied to anomaly-based network intrusion detection system (NIDS) to detect malicious traffic. To expand the usage scenarios of DL-based methods, federated learning (FL) allows multiple users to train a global model on the basis of respecting individual data privacy. However, it has not yet been systematically evaluated how robust FL-based NIDSs are against existing privacy attacks under existing defenses. To address this issue, we propose two privacy evaluation metrics designed for FL-based NIDSs, including (1) privacy score that evaluates the similarity between the original and recovered traffic features using reconstruction attacks, and (2) evasion rate against NIDSs using adversarial attack with the recovered traffic. We conduct experiments to illustrate that existing defenses provide little protection and the corresponding adversarial traffic can even evade the SOTA NIDS Kitsune. To defend against such attacks and build a more robust FL-based NIDS, we further propose FedDef, a novel optimization-based input perturbation defense strategy with theoretical guarantee. It achieves both high utility by minimizing the gradient distance and strong privacy protection by maximizing the input distance. We experimentally evaluate four existing defenses on four datasets and show that our defense outperforms all the baselines in terms of privacy protection with up to 7 times higher privacy score, while maintaining model accuracy loss within 3% under optimal parameter combination.

Yanpeng Sun, Jiahui Chen, Shan Zhang et al.

In this paper, we propose a novel Visual Reference Prompt (VRP) encoder that empowers the Segment Anything Model (SAM) to utilize annotated reference images as prompts for segmentation, creating the VRP-SAM model. In essence, VRP-SAM can utilize annotated reference images to comprehend specific objects and perform segmentation of specific objects in target image. It is note that the VRP encoder can support a variety of annotation formats for reference images, including \textbf{point}, \textbf{box}, \textbf{scribble}, and \textbf{mask}. VRP-SAM achieves a breakthrough within the SAM framework by extending its versatility and applicability while preserving SAM's inherent strengths, thus enhancing user-friendliness. To enhance the generalization ability of VRP-SAM, the VRP encoder adopts a meta-learning strategy. To validate the effectiveness of VRP-SAM, we conducted extensive empirical studies on the Pascal and COCO datasets. Remarkably, VRP-SAM achieved state-of-the-art performance in visual reference segmentation with minimal learnable parameters. Furthermore, VRP-SAM demonstrates strong generalization capabilities, allowing it to perform segmentation of unseen objects and enabling cross-domain segmentation. The source code and models will be available at https://github.com/syp2ysy/VRP-SAM

Jiahui Chen, Zehang Deng, Zeyu Zhang et al.

Deep learning-based image watermarking, while robust against conventional distortions, remains vulnerable to advanced adversarial and regeneration attacks. Conventional countermeasures, which jointly optimize the encoder and decoder via a noise layer, face 2 inevitable challenges: (1) decrease of clean accuracy due to decoder adversarial training and (2) limited robustness due to simultaneous training of all three advanced attacks. To overcome these issues, we propose AdvMark, a novel two-stage fine-tuning framework that decouples the defense strategies. In stage 1, we address adversarial vulnerability via a tailored adversarial training paradigm that primarily fine-tunes the encoder while only conditionally updating the decoder. This approach learns to move the image into a non-attackable region, rather than modifying the decision boundary, thus preserving clean accuracy. In stage 2, we tackle distortion and regeneration attacks via direct image optimization. To preserve the adversarial robustness gained in stage 1, we formulate a principled, constrained image loss with theoretical guarantees, which balances the deviation from cover and previous encoded images. We also propose a quality-aware early-stop to further guarantee the lower bound of visual quality. Extensive experiments demonstrate AdvMark outperforms with the highest image quality and comprehensive robustness, i.e. up to 29\%, 33\% and 46\% accuracy improvement for distortion, regeneration and adversarial attacks, respectively.

Philippe Hansen-Estruch, Jiahui Chen, Vivek Ramanujan et al.

Vision Transformer (ViT) autoencoders have emerged as compelling tokenizers for images, offering improved reconstruction over convolutional tokenizers. However, existing ViT tokenizers cannot explore this landscape as performance degrades outside training resolutions, and reliance on adversarial losses prevents stable scaling. ViTok (Hansen-Estruch et al., 2025) found that the compression ratio r mediates a reconstruction-generation trade-off where lower r means better reconstructions but harder generations, so improving tokenizer reconstruction is key to more Pareto-optimal tokenizers. We introduce ViTok-v2, which addresses these limitations with native resolution support via NaFlex for generalization across resolutions and aspect ratios, and a novel DINOv3 perceptual loss that replaces both LPIPS and GAN objectives for stable training at any scale. ViTok-v2 is trained on about 2B images and scaled to 5B parameters, the largest image autoencoder to date. ViTok-v2 matches or exceeds state-of-the-art reconstruction at 256p and outperforms all baselines at 512p and above. In joint scaling experiments with flow matching generators, we show that scaling both the autoencoder and the generator advances the Pareto frontier of this trade-off.

Lianchen Jia, Chaoyang Li, Ziqi Yuan et al.

Over the past decade, adaptive video streaming technology has witnessed significant advancements, particularly driven by the rapid evolution of deep learning techniques. However, the black-box nature of deep learning algorithms presents challenges for developers in understanding decision-making processes and optimizing for specific application scenarios. Although existing research has enhanced algorithm interpretability through decision tree conversion, interpretability does not directly equate to developers' subjective comprehensibility. To address this challenge, we introduce \texttt{ComTree}, the first bitrate adaptation algorithm generation framework that considers comprehensibility. The framework initially generates the complete set of decision trees that meet performance requirements, then leverages large language models to evaluate these trees for developer comprehensibility, ultimately selecting solutions that best facilitate human understanding and enhancement. Experimental results demonstrate that \texttt{ComTree} significantly improves comprehensibility while maintaining competitive performance, showing potential for further advancement. The source code is available at https://github.com/thu-media/ComTree.

Huan Yang, Jiahui Chen, Chaofan Ding et al.

Gestures are pivotal in enhancing co-speech communication. While recent works have mostly focused on point-level motion transformation or fully supervised motion representations through data-driven approaches, we explore the representation of gestures in co-speech, with a focus on self-supervised representation and pixel-level motion deviation, utilizing a diffusion model which incorporates latent motion features. Our approach leverages self-supervised deviation in latent representation to facilitate hand gestures generation, which are crucial for generating realistic gesture videos. Results of our first experiment demonstrate that our method enhances the quality of generated videos, with an improvement from 2.7 to 4.5% for FGD, DIV, and FVD, and 8.1% for PSNR, 2.5% for SSIM over the current state-of-the-art methods.

Qiang Chen, Xiangbo Su, Xinyu Zhang et al.

In this paper, we present a light-weight detection transformer, LW-DETR, which outperforms YOLOs for real-time object detection. The architecture is a simple stack of a ViT encoder, a projector, and a shallow DETR decoder. Our approach leverages recent advanced techniques, such as training-effective techniques, e.g., improved loss and pretraining, and interleaved window and global attentions for reducing the ViT encoder complexity. We improve the ViT encoder by aggregating multi-level feature maps, and the intermediate and final feature maps in the ViT encoder, forming richer feature maps, and introduce window-major feature map organization for improving the efficiency of interleaved attention computation. Experimental results demonstrate that the proposed approach is superior over existing real-time detectors, e.g., YOLO and its variants, on COCO and other benchmark datasets. Code and models are available at (https://github.com/Atten4Vis/LW-DETR).

Ruochi Zhang, Haoran Wu, Chang Liu et al.

Recent advances in protein language models have catalyzed significant progress in peptide sequence representation. Despite extensive exploration in this field, pre-trained models tailored for peptide-specific needs remain largely unaddressed due to the difficulty in capturing the complex and sometimes unstable structures of peptides. This study introduces a novel multi-view contrastive learning framework PepHarmony for the sequence-based peptide encoding task. PepHarmony innovatively combines both sequence- and structure-level information into a sequence-level encoding module through contrastive learning. We carefully select datasets from the Protein Data Bank (PDB) and AlphaFold database to encompass a broad spectrum of peptide sequences and structures. The experimental data highlights PepHarmony's exceptional capability in capturing the intricate relationship between peptide sequences and structures compared with the baseline and fine-tuned models. The robustness of our model is confirmed through extensive ablation studies, which emphasize the crucial roles of contrastive loss and strategic data sorting in enhancing predictive performance. The proposed PepHarmony framework serves as a notable contribution to peptide representations, and offers valuable insights for future applications in peptide drug discovery and peptide engineering. We have made all the source code utilized in this study publicly accessible via GitHub at https://github.com/zhangruochi/PepHarmony or http://www.healthinformaticslab.org/supp/.

Jiahui Chen, Weida Wang, Runhua Shi et al.

While diffusion model for audio-driven avatar video generation have achieved notable process in synthesizing long sequences with natural audio-visual synchronization and identity consistency, the generation of music-performance videos with camera motions remains largely unexplored. We present YingVideo-MV, the first cascaded framework for music-driven long-video generation. Our approach integrates audio semantic analysis, an interpretable shot planning module (MV-Director), temporal-aware diffusion Transformer architectures, and long-sequence consistency modeling to enable automatic synthesis of high-quality music performance videos from audio signals. We construct a large-scale Music-in-the-Wild Dataset by collecting web data to support the achievement of diverse, high-quality results. Observing that existing long-video generation methods lack explicit camera motion control, we introduce a camera adapter module that embeds camera poses into latent noise. To enhance continulity between clips during long-sequence inference, we further propose a time-aware dynamic window range strategy that adaptively adjust denoising ranges based on audio embedding. Comprehensive benchmark tests demonstrate that YingVideo-MV achieves outstanding performance in generating coherent and expressive music videos, and enables precise music-motion-camera synchronization. More videos are available in our project page: https://giantailab.github.io/YingVideo-MV/ .

Zheng Jiang, Yiming Chen, Nan He et al.

Recent multimodal large language models (MLLMs) have begun to support Thinking with Images by invoking visual tools such as zooming and cropping during inference. Yet these systems remain brittle in fine-grained visual reasoning because they must decide where to look before they have access to the evidence needed to make that decision correctly. We identify this circular dependency as the Grounding Paradox. To address it, we propose Test-Time Scaling over Perception (TTSP), a framework that treats perception itself as a scalable inference process. TTSP generates multiple exploratory perception traces, filters unreliable traces using entropy-based confidence estimation, distills validated observations into structured knowledge, and iteratively refines subsequent exploration toward unresolved uncertainty. Extensive experiments on high-resolution and general multimodal reasoning benchmarks show that TTSP consistently outperforms strong baselines across backbone sizes, while also exhibiting favorable scalability and token efficiency. Our results suggest that scaling perception at test time is a promising direction for robust multimodal reasoning under perceptual uncertainty.

Yunyue Su, Jiahui Chen, Zao Jiang et al.

Structure elucidation is a fundamental technique for understanding the microscopic composition of matter and is widely applied across various disciplines in the natural sciences and engineering. However, existing methods often rely heavily on prior databases or known structural information, making it difficult to resolve unknown structures. In addition, complex structures typically require the joint analysis of multiple spectroscopic modalities. This process heavily depends on expert domain knowledge and is often accompanied by high costs in terms of both time and instrumentation. To address these challenges, we propose SpectraLLM, the first large language model designed to support multi-modal spectroscopic joint reasoning. SpectraLLM is capable of processing either single or multiple spectroscopic inputs and performing end-to-end structure elucidation. By integrating continuous and discrete spectroscopic modalities into a shared semantic space, SpectraLLM learns to uncover substructural patterns that are consistent and complementary across spectra, enabling precise molecular structure elucidation. We pretrain and fine-tune SpectraLLM in the domain of small molecules, and evaluate it on six standardized, publicly available chemical datasets. The model achieves state-of-the-art performance, significantly outperforming existing approaches trained on single modalities. Notably, SpectraLLM demonstrates strong robustness and generalization even for single-spectrum inference, while its multi-modal reasoning capability further improves the accuracy of structural prediction.

Jiahui Chen, Lei Fu, Jian Cui et al.

Large-scale Text-to-SQL benchmarks such as BIRD typically assume complete and accurate database annotations as well as readily available external knowledge, which fails to reflect common industrial settings where annotations are missing, incomplete, or erroneous. This mismatch substantially limits the real-world applicability of state-of-the-art (SOTA) Text-to-SQL systems. To bridge this gap, we explore a database-centric approach that leverages intrinsic, fine-grained information residing in relational databases to construct missing evidence and improve Text-to-SQL accuracy under annotation-scarce conditions. Our key hypothesis is that when a query requires multi-step reasoning over extensive table information, existing methods often struggle to reliably identify and utilize the truly relevant knowledge. We therefore propose to "cache" query-relevant knowledge on the database side in advance, so that it can be selectively activated at inference time. Based on this idea, we introduce Companion Agents (CA), a new Text-to-SQL paradigm that incorporates a group of agents accompanying database schemas to proactively mine and consolidate hidden inter-table relations, value-domain distributions, statistical regularities, and latent semantic cues before query generation. Experiments on BIRD under the fully missing evidence setting show that CA recovers +4.49 / +4.37 / +14.13 execution accuracy points on RSL-SQL / CHESS / DAIL-SQL, respectively, with larger gains on the Challenging subset +9.65 / +7.58 / +16.71. These improvements stem from CA's automatic database-side mining and evidence construction, suggesting a practical path toward industrial-grade Text-to-SQL deployment without reliance on human-curated evidence.

Jiahui Chen, Philippe Hansen-Estruch, Xiaochuang Han et al.

Unified multimodal generation architectures that jointly produce text and images have recently emerged as a promising direction for text-to-image (T2I) synthesis. However, many existing systems rely on explicit modality switching, generating reasoning text before switching manually to image generation. This separate, sequential inference process limits cross-modal coupling and prohibits automatic multimodal generation. This work explores post-training to achieve fully unified text-image generation, where models autonomously transition from textual reasoning to visual synthesis within a single inference process. We examine the impact of joint text-image generation on T2I performance and the relative importance of each modality during post-training. We additionally explore different post-training data strategies, showing that a targeted dataset addressing specific limitations achieves superior results compared to broad image-caption corpora or benchmark-aligned data. Using offline, reward-weighted post-training with fully self-generated synthetic data, our approach enables improvements in multimodal image generation across four diverse T2I benchmarks, demonstrating the effectiveness of reward-weighting both modalities and strategically designed post-training data.

Jiahui Chen, Bo Peng, Lianchen Jia et al.

Content-aware streaming requires dynamic, chunk-level importance weights to optimize subjective quality of experience (QoE). However, direct human annotation is prohibitively expensive while vision-saliency models generalize poorly. We introduce HiVid, the first framework to leverage Large Language Models (LLMs) as a scalable human proxy to generate high-fidelity weights for both Video-on-Demand (VOD) and live streaming. We address 3 non-trivial challenges: (1) To extend LLMs' limited modality and circumvent token limits, we propose a perception module to assess frames in a local context window, autoregressively building a coherent understanding of the video. (2) For VOD with rating inconsistency across local windows, we propose a ranking module to perform global re-ranking with a novel LLM-guided merge-sort algorithm. (3) For live streaming which requires low-latency, online inference without future knowledge, we propose a prediction module to predict future weights with a multi-modal time series model, which comprises a content-aware attention and adaptive horizon to accommodate asynchronous LLM inference. Extensive experiments show HiVid improves weight prediction accuracy by up to 11.5\% for VOD and 26\% for live streaming over SOTA baselines. Real-world user study validates HiVid boosts streaming QoE correlation by 14.7\%.

Ying Wang, Jiahui Chen, Dejun Jiang

Storage systems are fundamental to modern computing infrastructures, yet ensuring their correctness remains challenging in practice. Despite decades of research on system testing, many storage-system failures (including durability, ordering, recovery, and consistency violations) remain difficult to expose systematically. This difficulty stems not primarily from insufficient testing tooling, but from intrinsic properties of storage-system execution, including nondeterministic interleavings, long-horizon state evolution, and correctness semantics that span multiple layers and execution phases. This survey adopts a storage-centric view of system testing and organizes existing techniques according to the execution properties and failure mechanisms they target. We review a broad spectrum of approaches, ranging from concurrency testing and long-running workloads to crash-consistency analysis, hardware-level semantic validation, and distributed fault injection, and analyze their fundamental strengths and limitations. Within this framework, we examine fuzzing as an automated testing paradigm, highlighting systematic mismatches between conventional fuzzing assumptions and storage-system semantics, and discuss how recent artificial intelligence advances may complement fuzzing through state-aware and semantic guidance. Overall, this survey provides a unified perspective on storage-system correctness testing and outlines key challenges

Ao Xie, Jiahui Chen, Quanzhi Zhu et al.

Dense retrieval has become a foundational paradigm in modern search systems, especially on short-video platforms. However, most industrial systems adopt a self-reinforcing training pipeline that relies on historically exposed user interactions for supervision. This paradigm inevitably leads to a filter bubble effect, where potentially relevant but previously unseen content is excluded from the training signal, biasing the model toward narrow and conservative retrieval. In this paper, we present CroPS (Cross-Perspective Positive Samples), a novel retrieval data engine designed to alleviate this problem by introducing diverse and semantically meaningful positive examples from multiple perspectives. CroPS enhances training with positive signals derived from user query reformulation behavior (query-level), engagement data in recommendation streams (system-level), and world knowledge synthesized by large language models (knowledge-level). To effectively utilize these heterogeneous signals, we introduce a Hierarchical Label Assignment (HLA) strategy and a corresponding H-InfoNCE loss that together enable fine-grained, relevance-aware optimization. Extensive experiments conducted on Kuaishou Search, a large-scale commercial short-video search platform, demonstrate that CroPS significantly outperforms strong baselines both offline and in live A/B tests, achieving superior retrieval performance and reducing query reformulation rates. CroPS is now fully deployed in Kuaishou Search, serving hundreds of millions of users daily.

Xinyuan Liu, Jiahui Chen, Bocheng Hu et al.

Data quality remains an important challenge in data-driven systems, as errors in tabular data can severely compromise downstream analytics and machine learning performance. Although numerous error detection algorithms have been proposed, the lack of diverse, real-world error datasets limits comprehensive evaluation. Manual error annotation is both time-consuming and inconsistent, motivating the exploration of synthetic error generation as an alternative. In this work, we introduce TableEG, a framework that leverages large language models (LLMs) to generate authentic errors. By employing a table fine-tuning strategy and a triplet representation $(I, T, O)$ to model error generation, detection, and correction tasks, TableEG captures the complex dependencies inherent in two-dimensional tables. Trained on 12 real-world datasets spanning 10 diverse domains, TableEG ensures that the synthesized errors faithfully reflect authentic error distributions. Experimental results indicate that errors generated by TableEG exhibit superior pattern and distribution similarity compared to both rule-based methods and LLM-generated errors without fine-tuning. Furthermore, performance metrics on TableEG-generated errors closely align with those on real-world errors across nearly all datasets and detection algorithms, particularly for machine learning based detection techniques. Overall, TableEG not only bridges the gap between synthetic and real-world errors but also establishes a robust benchmark for subsequent error detection and correction tasks.

Jiahui Chen, Candace Ross, Reyhane Askari-Hemmat et al.

The steady improvements of text-to-image (T2I) generative models lead to slow deprecation of automatic evaluation benchmarks that rely on static datasets, motivating researchers to seek alternative ways to evaluate the T2I progress. In this paper, we explore the potential of multi-modal large language models (MLLMs) as evaluator agents that interact with a T2I model, with the objective of assessing prompt-generation consistency and image aesthetics. We present Multimodal Text-to-Image Eval (MT2IE), an evaluation framework that iteratively generates prompts for evaluation, scores generated images and matches T2I evaluation of existing benchmarks with a fraction of the prompts used in existing static benchmarks. Moreover, we show that MT2IE's prompt-generation consistency scores have higher correlation with human judgment than scores previously introduced in the literature. MT2IE generates prompts that are efficient at probing T2I model performance, producing the same relative T2I model rankings as existing benchmarks while using only 1/80th the number of prompts for evaluation.

Jiahui Chen, Yang Huan, Runhua Shi et al.

Gestures are essential for enhancing co-speech communication, offering visual emphasis and complementing verbal interactions. While prior work has concentrated on point-level motion or fully supervised data-driven methods, we focus on co-speech gestures, advocating for weakly supervised learning and pixel-level motion deviations. We introduce a weakly supervised framework that learns latent representation deviations, tailored for co-speech gesture video generation. Our approach employs a diffusion model to integrate latent motion features, enabling more precise and nuanced gesture representation. By leveraging weakly supervised deviations in latent space, we effectively generate hand gestures and mouth movements, crucial for realistic video production. Experiments show our method significantly improves video quality, surpassing current state-of-the-art techniques.

Jiahui Chen, Amy Zhang, Adriana Romero-Soriano

Image generation abilities of text-to-image diffusion models have significantly advanced, yielding highly photo-realistic images from descriptive text and increasing the viability of leveraging synthetic images to train computer vision models. To serve as effective training data, generated images must be highly realistic while also sufficiently diverse within the support of the target data distribution. Yet, state-of-the-art conditional image generation models have been primarily optimized for creative applications, prioritizing image realism and prompt adherence over conditional diversity. In this paper, we investigate how to improve the diversity of generated images with the goal of increasing their effectiveness to train downstream image classification models, without fine-tuning the image generation model. We find that conditioning the generation process on an augmented real image and text prompt produces generations that serve as effective synthetic datasets for downstream training. Conditioning on real training images contextualizes the generation process to produce images that are in-domain with the real image distribution, while data augmentations introduce visual diversity that improves the performance of the downstream classifier. We validate augmentation-conditioning on a total of five established long-tail and few-shot image classification benchmarks and show that leveraging augmentations to condition the generation process results in consistent improvements over the state-of-the-art on the long-tailed benchmark and remarkable gains in extreme few-shot regimes of the remaining four benchmarks. These results constitute an important step towards effectively leveraging synthetic data for downstream training.

Chao Huang, Jiahui Chen, Chen Chen et al.

The design of crystal materials plays a critical role in areas such as new energy development, biomedical engineering, and semiconductors. Recent advances in data-driven methods have enabled the generation of diverse crystal structures. However, most existing approaches still rely on random sampling without strict constraints, requiring multiple post-processing steps to identify stable candidates with the desired physical and chemical properties. In this work, we present a new constrained generation framework that takes multiple constraints as input and enables the generation of crystal structures with specific chemical and properties. In this framework, intermediate constraints, such as symmetry information and composition ratio, are generated by a constraint generator based on large language models (LLMs), which considers the target properties. These constraints are then used by a subsequent crystal structure generator to ensure that the structure generation process is under control. Our method generates crystal structures with a probability of meeting the target properties that is more than twice that of existing approaches. Furthermore, nearly 100% of the generated crystals strictly adhere to predefined chemical composition, eliminating the risks of supply chain during production.

Wensheng Gan, Lili Chen, Shicheng Wan et al.

Analyzing sequence data usually leads to the discovery of interesting patterns and then anomaly detection. In recent years, numerous frameworks and methods have been proposed to discover interesting patterns in sequence data as well as detect anomalous behavior. However, existing algorithms mainly focus on frequency-driven analytic, and they are challenging to be applied in real-world settings. In this work, we present a new anomaly detection framework called DUOS that enables Discovery of Utility-aware Outlier Sequential rules from a set of sequences. In this pattern-based anomaly detection algorithm, we incorporate both the anomalousness and utility of a group, and then introduce the concept of utility-aware outlier sequential rule (UOSR). We show that this is a more meaningful way for detecting anomalies. Besides, we propose some efficient pruning strategies w.r.t. upper bounds for mining UOSR, as well as the outlier detection. An extensive experimental study conducted on several real-world datasets shows that the proposed DUOS algorithm has a better effectiveness and efficiency. Finally, DUOS outperforms the baseline algorithm and has a suitable scalability.

Jinbao Miao, Shicheng Wan, Wensheng Gan et al.

Traditional high-utility itemset mining (HUIM) aims to determine all high-utility itemsets (HUIs) that satisfy the minimum utility threshold (\textit{minUtil}) in transaction databases. However, in most applications, not all HUIs are interesting because only specific parts are required. Thus, targeted mining based on user preferences is more important than traditional mining tasks. This paper is the first to propose a target-based HUIM problem and to provide a clear formulation of the targeted utility mining task in a quantitative transaction database. A tree-based algorithm known as Target-based high-Utility iteMset querying using (TargetUM) is proposed. The algorithm uses a lexicographic querying tree and three effective pruning strategies to improve the mining efficiency. We implemented experimental validation on several real and synthetic databases, and the results demonstrate that the performance of \textbf{TargetUM} is satisfactory, complete, and correct. Finally, owing to the lexicographic querying tree, the database no longer needs to be scanned repeatedly for multiple queries.

Chao Huang, Jiahui Chen, Lianghao Xia et al.

Session-based recommendation plays a central role in a wide spectrum of online applications, ranging from e-commerce to online advertising services. However, the majority of existing session-based recommendation techniques (e.g., attention-based recurrent network or graph neural network) are not well-designed for capturing the complex transition dynamics exhibited with temporally-ordered and multi-level inter-dependent relation structures. These methods largely overlook the relation hierarchy of item transitional patterns. In this paper, we propose a multi-task learning framework with Multi-level Transition Dynamics (MTD), which enables the jointly learning of intra- and inter-session item transition dynamics in automatic and hierarchical manner. Towards this end, we first develop a position-aware attention mechanism to learn item transitional regularities within individual session. Then, a graph-structured hierarchical relation encoder is proposed to explicitly capture the cross-session item transitions in the form of high-order connectivities by performing embedding propagation with the global graph context. The learning process of intra- and inter-session transition dynamics are integrated, to preserve the underlying low- and high-level item relationships in a common latent space. Extensive experiments on three real-world datasets demonstrate the superiority of MTD as compared to state-of-the-art baselines.

Jiahui Chen, Weihua Geng, Guo-Wei Wei

Monte Carlo (MC) methods are important computational tools for molecular structure optimizations and predictions. When solvent effects are explicitly considered, MC methods become very expensive due to the large degree of freedom associated with the water molecules and mobile ions. Alternatively implicit-solvent MC can largely reduce the computational cost by applying a mean field approximation to solvent effects and meanwhile maintains the atomic detail of the target molecule. The two most popular implicit-solvent models are the Poisson-Boltzmann (PB) model and the Generalized Born (GB) model in a way such that the GB model is an approximation to the PB model but is much faster in simulation time. In this work, we develop a machine learning-based implicit-solvent Monte Carlo (MLIMC) method by combining the advantages of both implicit solvent models in accuracy and efficiency. Specifically, the MLIMC method uses a fast and accurate PB-based machine learning (PBML) scheme to compute the electrostatic solvation free energy at each step. We validate our MLIMC method by using a benzene-water system and a protein-water system. We show that the proposed MLIMC method has great advantages in speed and accuracy for molecular structure optimization and prediction.

Jiahui Chen, Joe Breen, Jeff M. Phillips et al.

Network traffic classification that is widely applicable and highly accurate is valuable for many network security and management tasks. A flexible and easily configurable classification framework is ideal, as it can be customized for use in a wide variety of networks. In this paper, we propose a highly configurable and flexible machine learning traffic classification method that relies only on statistics of sequences of packets to distinguish known, or approved, traffic from unknown traffic. Our method is based on likelihood estimation, provides a measure of certainty for classification decisions, and can classify traffic at adjustable certainty levels. Our classification method can also be applied in different classification scenarios, each prioritizing a different classification goal. We demonstrate how our classification scheme and all its configurations perform well on real-world traffic from a high performance computing network environment.

Jiahui Chen, Wensheng Gan, Muchuang Hu et al.

Owing to some special characteristics and features, blockchain is a very useful technique that can securely organize diverse devices in a smart city. It finds wide applications, especially in distributed environments, where entities such as wireless sensors need to be certain of the authenticity of the server. As contemporary blockchain techniques that address post-quantum concerns have not been designed, in this study, we investigate a blockchain in the post-quantum setting and seek to discover how it can resist attacks from quantum computing. In addition, traditional proof of work (PoW)-based consensus protocols such as Bitcoin cannot supply memory mining, and the transaction capacity of each block in a blockchain is limited and needs to be expanded. Thus, a new post-quantum proof of work (post-quantum PoW) consensus algorithm for security and privacy of smart city applications is proposed. It can be used to not only protect a blockchain under a quantum computing attack compared to existing classical hash-based PoW algorithms but also to supply memory mining. Meanwhile, an identity-based post-quantum signature is embedded into a transaction process to construct lightweight transactions. Subsequently, we provide a detailed description on the execution of the post-quantum lightweight transaction in a blockchain. Overall, this work can help enrich the research on future post-quantum blockchain and support the construction or architecture of emerging blockchain-based smart cities.

Muchuang Hu, Jiahui Chen, Wensheng Gan et al.

Mining is the important part of the blockchain used the proof of work (PoW) on its consensus, looking for the matching block through testing a number of hash calculations. In order to attract more hash computing power, the miner who finds the proper block can obtain some rewards. Actually, these hash calculations ensure that the data of the blockchain is not easily tampered. Thus, the incentive mechanism for mining affects the security of the blockchain directly. This paper presents an approach to attack against the difficulty adjustment algorithm (abbreviated as DAA) used in blockchain mining, which has a direct impact on miners' earnings. In this method, the attack miner jumps between different blockchains to get more benefits than the honest miner who keep mining on only one blockchain. We build a probabilistic model to simulate the time to obtain the next block at different hash computing power called hashrate. Based on this model, we analyze the DAAs of the major cryptocurrencies, including Bitcoin, Bitcoin Cash, Zcash, and Bitcoin Gold. We further verify the effectiveness of this attack called jumping mining through simulation experiments, and also get the characters for the attack in the public block data of Bitcoin Gold. Finally, we give an improved DAA scheme against this attack. Extensive experiments are provided to support the efficiency of our designed scheme.

Xinyu Wei, Jiahui Chen, Jing Chen et al.

In the era of information explosion, facing complex information, it is difficult for users to choose the information of interest, and businesses also need detailed information on ways to let the ad stand out. By this time, it is recommended that a good way. We firstly by using random interviews, simulations, asking experts, summarizes methods outlined the main factors affecting the scores of books that users drew. In order to further illustrate the impact of these factors, we also by combining the AHP consistency test, then fuzzy evaluation method, empowered each factor, influencing factors and the degree of influence come. For the second question, predict user evaluation of the listed books from the predict annex. First, given the books Annex labels, user data extraction scorebooks and mathematical analysis of data obtained from SPSS user preferences and then use software to nearest neighbor analysis to result in predicted value.