arXiv subjects

Yuxiang Zeng

Publications and source records attributed to Yuxiang Zeng.

12 recordsLinked to original sources

arXiv2026

JAVEDIT: Joint Audio-Visual Instruction-Guided Video Editing with Agentic Data Curation

While instruction-based video editing has seen significant progress, joint audio-visual editing remains constrained by the absence of dedicated datasets and benchmarks. To bridge this gap, we present JAVEdit-100k, the first large-scale, high-quality dataset tailored for instruction-guided joint audio-visual editing. Focusing on human-centric videos, JAVEdit-100k comprises approximately 100K editing triplets spanning five distinct categories, including subject editing and speech editing. This dataset is rigorously constructed via four meticulously designed generation pipelines, seamlessly paired with an agent-in-the-loop quality control mechanism. Furthermore, to address the lack of standardized evaluation within the field, we introduce JAVEditBench, a comprehensive benchmark featuring curated source videos and human-aligned instructions across all editing categories. Finally, we propose JAVEdit, a pioneering baseline model for instruction-guided joint audio-visual editing. Experiments show that \model\ outperforms all baselines on five of six evaluation metrics.

cs.CV↗

arXiv2026

Distance Comparison Operations Are Not Silver Bullets in Vector Similarity Search: A Benchmark Study on Their Merits and Limits

Distance Comparison Operations (DCOs), which decide whether the distance between a data vector and a query is within a threshold, are a critical performance bottleneck in vector similarity search. Recent DCO methods that avoid full-dimensional distance computations promise significant speedups, but their readiness for production vector database systems remains an open question. To address this, we conduct a comprehensive benchmark of 8 DCO algorithms across 10 datasets (with up to 100M vectors and 12,288 dimensions) and diverse hardware configurations (CPUs with/without SIMD, and GPUs). Our study reveals that these methods are not silver bullets: their efficiency is highly sensitive to data dimensionality, degrades under out-of-distribution queries, and is unstable across hardware. Yet, our evaluation also demonstrates often-overlooked merits: they can accelerate index construction and data updates. Despite these benefits, their unstable performance, which can be slower than a full-dimensional scan, leads us to conclude that recent algorithmic advancements in DCO are not yet ready for production deployment.

cs.DB↗

arXiv2026

Unified and Efficient Approach for Multi-Vector Similarity Search

Multi-Vector Similarity Search is essential for fine-grained semantic retrieval in many real-world applications, offering richer representations than traditional single-vector paradigms. Due to the lack of native multi-vector index, existing methods rely on a filter-and-refine framework built upon single-vector indexes. By treating token vectors within each multi-vector object in isolation and ignoring their correlations, these methods face an inherent dilemma: aggressive filtering sacrifices recall, while conservative filtering incurs prohibitive computational cost during refinement. To address this limitation, we propose MV-HNSW, the first native hierarchical graph index designed for multi-vector data. MV-HNSW introduces a novel edge-weight function that satisfies essential properties (symmetry, cardinality robustness, and query consistency) for graph-based indexing, an accelerated multi-vector similarity computation algorithm, and an augmented search strategy that dynamically discovers topologically disconnected yet relevant candidates. Extensive experiments on seven real-world datasets show that MV-HNSW achieves state-of-the-art search performance, maintaining over 90% recall while reducing search latency by up to 14.0$\times$ compared to existing methods.

cs.DB↗

arXiv2026

IVEBench: Modern Benchmark Suite for Instruction-Guided Video Editing Assessment

Instruction-guided video editing has emerged as a rapidly advancing research direction, offering new opportunities for intuitive content transformation while also posing significant challenges for systematic evaluation. Existing video editing benchmarks fail to support the evaluation of instruction-guided video editing adequately and further suffer from limited source diversity, narrow task coverage and incomplete evaluation metrics. To address the above limitations, we introduce IVEBench, a modern benchmark suite specifically designed for instruction-guided video editing assessment. IVEBench comprises a diverse database of 600 high-quality source videos, spanning seven semantic dimensions, and covering video lengths ranging from 32 to 1,024 frames. It further includes 8 categories of editing tasks with 35 subcategories, whose prompts are generated and refined through large language models and expert review. Crucially, IVEBench establishes a three-dimensional evaluation protocol encompassing video quality, instruction compliance and video fidelity, integrating both traditional metrics and multimodal large language model-based assessments. Extensive experiments demonstrate the effectiveness of IVEBench in benchmarking state-of-the-art instruction-guided video editing methods, showing its ability to provide comprehensive and human-aligned evaluation outcomes.

cs.CV↗

arXiv2025

QoSGMAA: A Robust Multi-Order Graph Attention and Adversarial Framework for Sparse QoS Prediction

With the rapid advancement of internet technologies, network services have become critical for delivering diverse and reliable applications to users. However, the exponential growth in the number of available services has resulted in many similar offerings, posing significant challenges in selecting optimal services. Predicting Quality of Service (QoS) accurately thus becomes a fundamental prerequisite for ensuring reliability and user satisfaction. However, existing QoS prediction methods often fail to capture rich contextual information and exhibit poor performance under extreme data sparsity and structural noise. To bridge this gap, we propose a novel architecture, QoSMGAA, specifically designed to enhance prediction accuracy in complex and noisy network service environments. QoSMGAA integrates a multi-order attention mechanism to aggregate extensive contextual data and predict missing QoS values effectively. Additionally, our method incorporates adversarial neural networks to perform autoregressive supervised learning based on transformed interaction matrices. To capture complex, higher-order interactions among users and services, we employ a discrete sampling technique leveraging the Gumbel-Softmax method to generate informative negative samples. Comprehensive experimental validation conducted on large-scale real-world datasets demonstrates that our proposed model significantly outperforms existing baseline methods, highlighting its strong potential for practical deployment in service selection and recommendation scenarios.

cs.LG↗

arXiv2024

How Good Are Multi-dimensional Learned Indices? An Experimental Survey

Efficient indexing is fundamental for multi-dimensional data management and analytics. An emerging tendency is to directly learn the storage layout of multi-dimensional data by simple machine learning models, yielding the concept of Learned Index. Compared with the conventional indices used for decades (e.g., kd-tree and R-tree variants), learned indices are empirically shown to be both space- and time-efficient on modern architectures. However, there lacks a comprehensive evaluation of existing multi-dimensional learned indices under a unified benchmark, which makes it difficult to decide the suitable index for specific data and queries and further prevents the deployment of learned indices in real application scenarios. In this paper, we present the first in-depth empirical study to answer the question of how good multi-dimensional learned indices are. Six recently published indices are evaluated under a unified experimental configuration including index implementation, datasets, query workloads, and evaluation metrics. We thoroughly investigate the evaluation results and discuss the findings that may provide insights for future learned index design.

cs.DB↗

arXiv2023

Efficient and Private Federated Trajectory Matching

Federated Trajectory Matching (FTM) is gaining increasing importance in big trajectory data analytics, supporting diverse applications such as public health, law enforcement, and emergency response. FTM retrieves trajectories that match with a query trajectory from a large-scale trajectory database, while safeguarding the privacy of trajectories in both the query and the database. A naive solution to FTM is to process the query through Secure Multi-Party Computation (SMC) across the entire database, which is inherently secure yet inevitably slow due to the massive secure operations. A promising acceleration strategy is to filter irrelevant trajectories from the database based on the query, thus reducing the SMC operations. However, a key challenge is how to publish the query in a way that both preserves privacy and enables efficient trajectory filtering. In this paper, we design GIST, a novel framework for efficient Federated Trajectory Matching. GIST is grounded in Geo-Indistinguishability, a privacy criterion dedicated to locations. It employs a new privacy mechanism for the query that facilitates efficient trajectory filtering. We theoretically prove the privacy guarantee of the mechanism and the accuracy of the filtering strategy of GIST. Extensive evaluations on five real datasets show that GIST is significantly faster and incurs up to 3 orders of magnitude lower communication cost than the state-of-the-arts.

cs.DB↗

arXiv2023

A Fast Insertion Operator for Ridesharing over Time-Dependent Road Networks

Ridesharing has become a promising travel mode recently due to the economic and social benefits. As an essential operator, "insertion operator" has been extensively studied over static road networks. When a new request appears, the insertion operator is used to find the optimal positions of a worker's current route to insert the origin and destination of this request and minimize the travel time of this worker. Previous works study how to conduct the insertion operation efficiently in static road networks, however, in reality, the route planning should be addressed by considering the dynamic traffic scenario (i.e., a time-dependent road network). Unfortunately, existing solutions to the insertion operator become in efficient under this setting. Thus, this paper studies the insertion operator over time-dependent road networks. Specially, to reduce the high time complexity $O(n^3)$ of existing solution, we calculate the compound travel time functions along the route to speed up the calculation of the travel time between vertex pairs belonging to the route, as a result time complexity of an insertion can be reduced to $O(n^2)$. Finally, we further improve the method to a linear-time insertion algorithm by showing that it only needs $O(1)$ time to find the best position of current route to insert the origin when linearly enumerating each possible position for the new request's destination. Evaluations on two real-world and large-scale datasets show that our methods can accelerate the existing insertion algorithm by up to 25 times.

cs.DB↗

arXiv2023

Querying Shortest Path on Large Time-Dependent Road Networks with Shortcuts

Querying the shortest path between two vertexes is a fundamental operation in a variety of applications, which has been extensively studied over static road networks. However, in reality, the travel costs of road segments evolve over time, and hence the road network can be modeled as a time-dependent graph. In this paper, we study the shortest path query over large-scale time-dependent road networks. Existing work focuses on a hierarchical partition structure, which makes the index construction and travel cost query inefficient. To improve the efficiency of such queries, we propose a novel index by decomposing a road network into a tree structure and selecting a set of shortcuts on the tree to speed up the query processing. Specifically, we first formally define a shortcut selection problem over the tree decomposition of the time-dependent road network. This problem, which is proven to be NP-hard, aims to select and build the most effective shortcut set. We first devise a dynamic programming method with exact results to solve the selection problem. To obtain the optimal shortcut set quickly, we design an approximation algorithm that guarantees a 0.5-approximation ratio. Based on the novel tree structure, we devise a shortcut-based algorithm to answer the shortest path query over time-dependent road networks. Finally, we conduct extensive performance studies using large-scale real-world road networks. The results demonstrate that our method can achieve better efficiency and scalability than the state-of-the-art method.

cs.DB↗

arXiv2023

Accurate and Efficient Trajectory-based Contact Tracing with Secure Computation and Geo-Indistinguishability

Contact tracing has been considered as an effective measure to limit the transmission of infectious disease such as COVID-19. Trajectory-based contact tracing compares the trajectories of users with the patients, and allows the tracing of both direct contacts and indirect contacts. Although trajectory data is widely considered as sensitive and personal data, there is limited research on how to securely compare trajectories of users and patients to conduct contact tracing with excellent accuracy, high efficiency, and strong privacy guarantee. Traditional Secure Multiparty Computation (MPC) techniques suffer from prohibitive running time, which prevents their adoption in large cities with millions of users. In this work, we propose a technical framework called ContactGuard to achieve accurate, efficient, and privacy-preserving trajectory-based contact tracing. It improves the efficiency of the MPC-based baseline by selecting only a small subset of locations of users to compare against the locations of the patients, with the assist of Geo-Indistinguishability, a differential privacy notion for Location-based services (LBS) systems. Extensive experiments demonstrate that ContactGuard runs up to 2.6$\times$ faster than the MPC baseline, with no sacrifice in terms of the accuracy of contact tracing.

cs.DB↗

arXiv2023

Efficient and Accurate Range Counting on Privacy-preserving Spatial Data Federation [Technical Report]

A spatial data federation is a collection of data owners (e.g., a consortium of taxi companies), and collectively it could provide better location-based services (LBS). For example, car-hailing services over a spatial data federation allow end users to easily pick the best offers. We focus on the range counting queries, which are primitive operations in spatial databases but received little attention in related research, especially considering the privacy requirements from data owners, who are reluctant to disclose their proprietary data. We propose a grouping-based technical framework named FedGroup, which groups data owners without compromising privacy, and achieves superior query accuracy (up to 50% improvement) as compared to directly applying existing privacy mechanisms achieving Differential Privacy (DP). Our experimental results also demonstrate that FedGroup runs orders-of-magnitude faster than traditional Secure Multiparty Computation (MPC) based method, and FedGroup even scales to millions of data owners, which is a common setting in the era of ubiquitous mobile devices.

cs.DB↗

arXiv2023

Towards Capacity-Aware Broker Matching: From Recommendation to Assignment

Online real estate platforms are gaining increasing popularity, where a central issue is to match brokers with clients for potential housing transactions. Mainstream platforms match brokers via top-k recommendation. Yet we observe through extensive data analysis that such top-k recommendation tends to overload the top brokers, which notably degrades their service quality. In this paper, we propose to avoid such overloading in broker matching via the paradigm shift from recommendation to assignment. To this end, we design learned assignment with contextual bandits (LACB), a data-driven capacity-aware assignment scheme for broker matching which estimates broker-specific workload capacity in an online fashion and assigns brokers to clients from a global perspective to maximize the overall service quality. Extensive evaluations on synthetic and real-world datasets from an industrial online real estate platform validate the efficiency and effectiveness of our solution.

cs.DB↗