arXiv subjects

Shiyu Shen

Publications and source records attributed to Shiyu Shen.

At least 19 recordsLinked to original sources

arXiv2026

Knowledge Is Not Static: Order-Aware Hypergraph RAG for Language Models

Retrieval-augmented generation (RAG) enhances large language models by grounding outputs in retrieved knowledge. However, existing RAG methods including graph- and hypergraph-based approaches treat retrieved evidence as an unordered set, implicitly assuming permutation invariance. This assumption is misaligned with many real-world reasoning tasks, where outcomes depend not only on which interactions occur, but also on the order in which they unfold. We propose Order-Aware Knowledge Hypergraph RAG (OKH-RAG), which treats order as a first-class structural property. OKH-RAG represents knowledge as higher-order interactions within a hypergraph augmented with precedence structure, and reformulates retrieval as sequence inference over hyperedges. Instead of selecting independent facts, it recovers coherent interaction trajectories that reflect underlying reasoning processes. A learned transition model infers precedence directly from data without requiring explicit temporal supervision. We evaluate OKH-RAG on order-sensitive question answering and explanation tasks, including tropical cyclone and port operation scenarios. OKH-RAG consistently outperforms permutation-invariant baselines, and ablations show that these gains arise specifically from modeling interaction order. These results highlight a key limitation of set-based retrieval: effective reasoning requires not only retrieving relevant evidence, but organizing it into structured sequences.

cs.CL↗

arXiv2026

VideoTIR: Accurate Understanding for Long Videos with Efficient Tool-Integrated Reasoning

Existing Multimodal Large Language Models (MLLMs) often suffer from hallucinations in long video understanding (LVU), primarily due to the imbalance between textual and visual tokens. Observing that MLLMs handle short visual inputs well, recent LVU works alleviate hallucinations by automatically parsing the vast visual data into manageable segments that can be effectively processed by MLLMs. SFT-based tool-calling methods can serve this purpose, but they typically require vast amounts of fine-grained, high-quality data and suffer from constrained tool-calling trajectories. We propose a novel VideoTIR that leverages Reinforcement Learning (RL) to encourage proper usage of comprehensive multi-level toolkits for efficient long video understanding. VideoTIR explores both Zero-RL and SFT cold-starting to enable MLLMs to retrieve and focus on meaningful video segments/images/regions, enhancing long video understanding both accurately and efficiently. To reduce redundant tool-calling, we propose Toolkit Action Grouped Policy Optimization (TAGPO), which enhances the efficiency of the calling process through stepwise reward assignment and reuse of failed rollouts. Additionally, we develop a sandbox-based trajectory synthesis framework to generate high-quality trajectories data. Extensive experiments on three long-video QA benchmarks demonstrate the effectiveness and efficiency of our method.

cs.CV↗

arXiv2025

Developing Fundamental Diagrams for Urban Air Mobility Traffic Based on Physical Experiments

Urban Air Mobility (UAM) is an emerging application of unmanned aerial vehicles that promises to reduce travel time and alleviate congestion in urban transportation systems. As drone density increases, UAM traffic is expected to experience congestion similar to that in ground traffic. However, the fundamental characteristics of UAM traffic, particularly under real-world operating conditions, remain largely unexplored. This study proposes a general framework for constructing the fundamental diagram (FD) of UAM traffic by integrating theoretical analysis with physical experiments. To the best of our knowledge, this is the first study to derive UAM FDs using real-world physical experiment data. On the theoretical side, we design two drone control laws for collision avoidance and develop simulation-based traffic generation methods to produce diverse UAM traffic scenarios. Based on Edie's definition, traffic flow theory is then applied with a near-stationary traffic condition filtering method to construct the FD. To account for real-world disturbances and modeling uncertainties, we further conduct physical experiments on a reduced-scale testbed using Bitcraze Crazyflie drones. Both simulation and physical experiment trajectory data are collected and organized into the UAMTra2Flow dataset, which is analyzed using the proposed framework. Preliminary results indicate that classical FD structures for ground transportation, especially the Underwood model, are applicable to UAM systems. Notably, FD curves obtained from physical experiments exhibit deviations from simulation-based results, highlighting the importance of experimental validation. Finally, results from the reduced-scale testbed are scaled to realistic operating conditions to provide practical insights for future UAM traffic systems. The dataset and code for this paper are publicly available at https://github.com/CATS-Lab/UAM-FD.

cs.RO↗

arXiv2025

Contrastive Heliophysical Image Pretraining for Solar Dynamics Observatory Records

Deep learning has revolutionized solar image analysis, yet most approaches train task-specific encoders from scratch or rely on natural-image pretraining that ignores the unique characteristics of Solar Dynamics Observatory (SDO) data. We introduce SolarCHIP, a family of contrastively pretrained visual backbones tailored to multi-instrument SDO observations. SolarCHIP addresses three key challenges in solar imaging: multimodal sensing across AIA and HMI instruments, weak inter-class separability due to slow temporal evolution, and strong intra-class variability with sparse activity signals. Our pretraining framework employs a multi-granularity contrastive objective that jointly aligns (1) global class tokens across co-temporal AIA-HMI pairs to enhance temporal discrimination, (2) local patch tokens at fixed spatial indices to enforce position-consistent, modality-invariant features, and (3) intra-sample patches across different spatial locations to preserve fine-grained spatial structure. We train both CNN- and Vision Transformer-based autoencoders and demonstrate their effectiveness on two downstream tasks: cross-modal translation between HMI and AIA passbands via ControlNet, and full-disk flare classification. Experimental results show that SolarCHIP achieves state-of-the-art performance across both tasks, with particularly strong gains in low-resource settings where labeled data is limited. Ablation studies confirm that each contrastive component contributes essential discriminative capacity at different granularities. By publicly releasing pretrained weights and training code, we provide the heliophysics community with a practical, plug-and-play feature extractor that reduces computational requirements, improves label efficiency, and establishes a reusable foundation for diverse solar imaging applications.

cs.CV↗

arXiv2025

HHEML: Hybrid Homomorphic Encryption for Privacy-Preserving Machine Learning on Edge

Privacy-preserving machine learning (PPML) is an emerging topic to handle secure machine learning inference over sensitive data in untrusted environments. Fully homomorphic encryption (FHE) enables computation directly on encrypted data on the server side, making it a promising approach for PPML. However, it introduces significant communication and computation overhead on the client side, making it impractical for edge devices. Hybrid homomorphic encryption (HHE) addresses this limitation by combining symmetric encryption (SE) with FHE to reduce the computational cost on the client side, and combining with an FHE-friendly SE can also lessen the processing overhead on the server side, making it a more balanced and efficient alternative. Our work proposes a hardware-accelerated HHE architecture built around a lightweight symmetric cipher optimized for FHE compatibility and implemented as a dedicated hardware accelerator. To the best of our knowledge, this is the first design to integrate an end-to-end HHE framework with hardware acceleration. Beyond this, we also present several microarchitectural optimizations to achieve higher performance and energy efficiency. The proposed work is integrated into a full PPML pipeline, enabling secure inference with significantly lower latency and power consumption than software implementations. Our contributions validate the feasibility of low-power, hardware- accelerated HHE for edge deployment and provide a hardware- software co-design methodology for building scalable, secure machine learning systems in resource-constrained environments. Experiments on a PYNQ-Z2 platform with the MNIST dataset show over a 50x reduction in client-side encryption latency and nearly a 2x gain in hardware throughput compared to existing FPGA-based HHE accelerators.

cs.CR↗

arXiv2025

Dynamic Random Bipartite Matching under Spatiotemporal Heterogeneity: General Models and Application to Mobility Services

This paper explores a variant of bipartite matching problem, referred to as the Spatiotemporal Random Bipartite Matching Problem (ST-RBMP), that accommodates randomness and heterogeneity in the spatial distributions and temporal arrivals of bipartite vertices. This type of problem can be applied to many location-based services, such as shared mobility systems, where randomly arriving customers and vehicles must be matched dynamically. This paper proposes a new modeling framework to address ST-RBMP's challenges associated with the spatiotemporal heterogeneity, dynamics, and stochastic decision-making. The objective is to dynamically determine the optimal vehicle/customer pooling intervals and maximum matching radii that minimize the system-wide matching costs, including customer and vehicle waiting times and matching distances. Closed-form formulas for estimating the expected matching distances under a maximum matching radius are developed for static and homogeneous RBMPs, and then extended to accommodate spatial heterogeneity via continuum approximation. The ST-RBMP is then formulated as an optimal control problem where optimal values of pooling intervals and matching radii are solved over time and space. A series of experiments with simulated data are conducted to demonstrate that the proposed formulas for static RBMPs under matching radius and spatial heterogeneity yield very accurate results on estimating matching probabilities and distances. Additional numerical results are presented to demonstrate the effectiveness of the proposed ST-RBMP modeling framework in designing dynamic matching strategies for mobility services under various demand and supply patterns, which offers key managerial insights for mobility service operators.

math.OC↗

arXiv2025

Preserving Domain Generalization in Fine-Tuning via Joint Parameter Selection

Domain generalization seeks to develop models trained on a limited set of source domains that are capable of generalizing effectively to unseen target domains. While the predominant approach leverages large-scale pre-trained vision models as initialization, recent studies have highlighted that full fine-tuning can compromise the intrinsic generalization capabilities of these models. To address this limitation, parameter-efficient adaptation strategies have emerged, wherein only a subset of model parameters is selectively fine-tuned, thereby balancing task adaptation with the preservation of generalization. Motivated by this paradigm, we introduce Joint Parameter Selection (JPS), a novel method that restricts updates to a small, sparse subset of parameters, thereby retaining and harnessing the generalization strength of pre-trained models. Theoretically, we establish a generalization error bound that explicitly accounts for the sparsity of parameter updates, thereby providing a principled justification for selective fine-tuning. Practically, we design a selection mechanism employing dual operators to identify and update parameters exhibiting consistent and significant gradients across all source domains. Extensive benchmark experiments demonstrate that JPS achieves superior performance compared to state-of-the-art domain generalization methods, substantiating both the efficiency and efficacy of the proposed approach.

cs.CV↗

arXiv2025

Towards Full-Scenario Safety Evaluation of Automated Vehicles: A Volume-Based Method

With the rapid development of automated vehicles (AVs) in recent years, commercially available AVs are increasingly demonstrating high-level automation capabilities. However, most existing AV safety evaluation methods are primarily designed for simple maneuvers such as car-following and lane-changing. While suitable for basic tests, these methods are insufficient for assessing high-level automation functions deployed in more complex environments. First, these methods typically use crash rate as the evaluation metric, whose accuracy heavily depends on the quality and completeness of naturalistic driving environment data used to estimate scenario probabilities. Such data is often difficult and expensive to collect. Second, when applied to diverse scenarios, these methods suffer from the curse of dimensionality, making large-scale evaluation computationally intractable. To address these challenges, this paper proposes a novel framework for full-scenario AV safety evaluation. A unified model is first introduced to standardize the representation of diverse driving scenarios. This modeling approach constrains the dimension of most scenarios to a regular highway setting with three lanes and six surrounding background vehicles, significantly reducing dimensionality. To further avoid the limitations of probability-based method, we propose a volume-based evaluation method that quantifies the proportion of risky scenarios within the entire scenario space. For car-following scenarios, we prove that the set of safe scenarios is convex under specific settings, enabling exact volume computation. Experimental results validate the effectiveness of the proposed volume-based method using both AV behavior models from existing literature and six production AV models calibrated from field-test trajectory data in the Ultra-AV dataset. Code and data will be made publicly available upon acceptance of this paper.

cs.RO↗

arXiv2025

Hyperspectral Image Generation with Unmixing Guided Diffusion Model

We address hyperspectral image (HSI) synthesis, a problem that has garnered growing interest yet remains constrained by the conditional generative paradigms that limit sample diversity. While diffusion models have emerged as a state-of-the-art solution for high-fidelity image generation, their direct extension from RGB to hyperspectral domains is challenged by the high spectral dimensionality and strict physical constraints inherent to HSIs. To overcome the challenges, we introduce a diffusion framework explicitly guided by hyperspectral unmixing. The approach integrates two collaborative components: (i) an unmixing autoencoder that projects generation from the image domain into a low-dimensional abundance manifold, thereby reducing computational burden while maintaining spectral fidelity; and (ii) an abundance diffusion process that enforces non-negativity and sum-to-one constraints, ensuring physical consistency of the synthesized data. We further propose two evaluation metrics tailored to hyperspectral characteristics. Comprehensive experiments, assessed with both conventional measures and the proposed metrics, demonstrate that our method produces HSIs with both high quality and diversity, advancing the state of the art in hyperspectral data generation.

cs.CV↗

arXiv2025

A Pretrained Probabilistic Transformer for City-Scale Traffic Volume Prediction

City-scale traffic volume prediction plays a pivotal role in intelligent transportation systems, yet remains a challenge due to the inherent incompleteness and bias in observational data. Although deep learning-based methods have shown considerable promise, most existing approaches produce deterministic point estimates, thereby neglecting the uncertainty arising from unobserved traffic flows. Furthermore, current models are typically trained in a city-specific manner, which hinders their generalizability and limits scalability across diverse urban contexts. To overcome these limitations, we introduce TrafficPPT, a Pretrained Probabilistic Transformer designed to model traffic volume as a distributional aggregation of trajectories. Our framework fuses heterogeneous data sources-including real-time observations, historical trajectory data, and road network topology-enabling robust and uncertainty-aware traffic inference. TrafficPPT is initially pretrained on large-scale simulated data spanning multiple urban scenarios, and later fine-tuned on target cities to ensure effective domain adaptation. Experiments on real-world datasets show that TrafficPPT consistently surpasses state-of-the-art baselines, particularly under conditions of extreme data sparsity. Code will be open.

cs.LG↗

arXiv2024

Self-control: A Better Conditional Mechanism for Masked Autoregressive Model

Autoregressive conditional image generation algorithms are capable of generating photorealistic images that are consistent with given textual or image conditions, and have great potential for a wide range of applications. Nevertheless, the majority of popular autoregressive image generation methods rely heavily on vector quantization, and the inherent discrete characteristic of codebook presents a considerable challenge to achieving high-quality image generation. To address this limitation, this paper introduces a novel conditional introduction network for continuous masked autoregressive models. The proposed self-control network serves to mitigate the negative impact of vector quantization on the quality of the generated images, while simultaneously enhancing the conditional control during the generation process. In particular, the self-control network is constructed upon a continuous mask autoregressive generative model, which incorporates multimodal conditional information, including text and images, into a unified autoregressive sequence in a serial manner. Through a self-attention mechanism, the network is capable of generating images that are controllable based on specific conditions. The self-control network discards the conventional cross-attention-based conditional fusion mechanism and effectively unifies the conditional and generative information within the same space, thereby facilitating more seamless learning and fusion of multimodal features.

cs.CV↗

arXiv2024

Average Distance of Random Bipartite Matching in One-dimensional Space and Networks

The bipartite matching problem is widely applied in the field of transportation; e.g., to find optimal matches between supply and demand over time and space. Recent efforts have been made on developing analytical formulas to estimate the expected matching distance in bipartite matching with randomly distributed vertices in two- or higher-dimensional spaces, but no accurate formulas currently exist for one-dimensional problems. This paper presents a set of closed-form formulas, without curve-fitting, that can provide accurate average distance estimates for one-dimensional random bipartite matching problems (RBMP). We first focus on a lattice case and propose a new method that relates the corresponding matching distance to the area size between a random walk path and the x-axis. This result directly leads to a straightforward closed-form formula for balanced RBMPs. For unbalanced RBMPs on a lattice, we first analyze the properties of an unbalanced random walk that can be related to balanced RBPMs after optimally removing a subset of unmatched points, and then derive a set of approximate formulas. Additionally, we build upon an optimal point removal strategy to derive a set of recursive formulas that can provide more accurate estimates. Then, we extend the results to three problem variants, including RBMPs with periodic boundaries, uniformly distributed points, and arbitrary-length line. Last, we shift our focus to regular networks, and use the one-dimensional results as building blocks to derive RBMP formulas. To verify the accuracy of the proposed formulas, a set of Monte-Carlo simulations are generated for a variety of matching problems settings. Results indicate that our proposed formulas provide quite accurate distance estimations for one-dimensional line segments and networks under a variety of conditions.

math.OC↗

arXiv2024

Expected Optimal Distances of Random Bipartite Matching in $D$-dimensional Spaces: Approximate Formulas and Applications to Mobility Services

Although many well-known algorithms can solve each bipartite matching problem instance efficiently, it remains an open question how one could estimate the expected optimal matching distance for arbitrary numbers of randomly distributed vertices in $D$-dimensional spaces (referred to as a random bipartite matching problem, or RBMP). This paper proposes a comprehensive modeling framework that yields closed-form approximate formulas for estimating the expected optimal matching cost across three interrelated but increasingly complex versions of RBMPs: (i) RBMP-I, where edge costs are independently and identically distributed (i.i.d.); (ii) RBMP-S, where edge costs represent distances between vertices uniformly distributed on the surface of a hyper-sphere in a $D$-dimensional Euclidean space; and (iii) RBMP-B, where the vertices are uniformly distributed in a hyper-ball within a $D$-dimensional L$^p$ metric space. A series of Monte-Carlo simulation experiments are conducted to verify the accuracy of the proposed formulas under varying parameter combinations. These proposed distance estimates could be key for strategic performance evaluation and resource planning in a wide variety of application contexts. As an illustration, we focus on on-demand mobility services (e.g., e-hailing taxi system). We show how the proposed distance formulas provide a theoretical foundation for the empirically assumed Cobb-Douglas matching function in the field, and reveal conditions under which it can work well. Our formulas can also be easily incorporated into optimization models to select on-demand mobility operation strategies (e.g., whether newly arriving customers shall be instantly matched or pooled into a batch for matching). Agent-based simulations are conducted to verify the predicted performance of the demand pooling strategy for two types of e-hailing taxi systems.

math.OC↗

arXiv2024

Domain Generalization Guided by Large-Scale Pre-Trained Priors

Domain generalization (DG) aims to train a model from limited source domains, allowing it to generalize to unknown target domains. Typically, DG models only employ large-scale pre-trained models during the initialization of fine-tuning. However, large-scale pre-trained models already possess the ability to resist domain shift. If we reference pre-trained models continuously during fine-tuning to maintain this ability, it could further enhance the generalization ability of the DG model. For this purpose, we introduce a new method called Fine-Tune with Large-scale pre-trained Priors (FT-LP), which incorporates the pre-trained model as a prior into the DG fine-tuning process, ensuring that the model refers to its pre-trained model at each optimization step. FT-LP comprises a theoretical framework and a simple implementation strategy. In theory, we verify the rationality of FT-LP by introducing a generalization error bound with the pre-trained priors for DG. In implementation, we utilize an encoder to simulate the model distribution, enabling the use of FT-LP when only pre-trained weights are available. In summary, we offer a new fine-tuning method for DG algorithms to utilize pre-trained models throughout the fine-tuning process. Through experiments on various datasets and DG models, our proposed method exhibits significant improvements, indicating its effectiveness.

cs.LG↗

arXiv2024

Mirror symmetry for parabolic Higgs bundles via $p$-adic integration

Applying the technique of $p$-adic integration, we prove the topological mirror symmetry conjecture of Hausel-Thaddeus for the moduli spaces of (strongly) parabolic Higgs bundles for the structure groups $\text{SL}_n$ and $\text{PGL}_n$, building on previous work of Groechenig-Wyss-Ziegler on the non-parabolic case. We also prove the $E$-polynomial of the smooth moduli space of parabolic $\text{GL}_n$-Higgs bundles is independent of the degree of the underlying vector bundles.

math.AG↗

arXiv2024

Tamely ramified geometric Langlands correspondence in positive characteristic

We prove a version of the tamely ramified geometric Langlands correspondence in positive characteristic for $GL_n(k)$. Let $k$ be an algebraically closed field of characteristic $p> n$. Let $X$ be a smooth projective curve over $k$ with marked points, and fix a parabolic subgroup of $GL_n(k)$ at each marked point. We denote by $\text{Bun}_{n,P}$ the moduli stack of (quasi-)parabolic vector bundles on $X$, and by $\mathcal{L}oc_{n,P}$ the moduli stack of parabolic flat connections such that the residue is nilpotent with respect to the parabolic reduction at each marked point. We construct an equivalence between the bounded derived category $D^{b}(\text{Qcoh}({\mathcal{L}oc_{n,P}^{0}}))$ of quasi-coherent sheaves on an open substack $\mathcal{L}oc_{n,P}^{0}\subset\mathcal{L}oc_{n,P}$, and the bounded derived category $D^{b}(\mathcal{D}^{0}_{\text{Bun}_{n,P}}\text{-mod})$ of $\mathcal{D}^{0}_{\text{Bun}_{n,P}}$-modules, where $\mathcal{D}^0_{\text{Bun}_{n,P}}$ is a localization of $\mathcal{D}_{\text{Bun}_{n,P}}$ the sheaf of crystalline differential operators on $\text{Bun}_{n,P}$. Thus we extend the work of Bezrukavnikov-Braverman to the tamely ramified case. We also prove a correspondence between flat connections on $X$ with regular singularities and meromorphic Higgs bundles on the Frobenius twist $X^{(1)}$ of $X$ with first order poles .

math.AG↗

arXiv2024

Be Bayesian by Attachments to Catch More Uncertainty

Bayesian Neural Networks (BNNs) have become one of the promising approaches for uncertainty estimation due to the solid theorical foundations. However, the performance of BNNs is affected by the ability of catching uncertainty. Instead of only seeking the distribution of neural network weights by in-distribution (ID) data, in this paper, we propose a new Bayesian Neural Network with an Attached structure (ABNN) to catch more uncertainty from out-of-distribution (OOD) data. We first construct a mathematical description for the uncertainty of OOD data according to the prior distribution, and then develop an attached Bayesian structure to integrate the uncertainty of OOD data into the backbone network. ABNN is composed of an expectation module and several distribution modules. The expectation module is a backbone deep network which focuses on the original task, and the distribution modules are mini Bayesian structures which serve as attachments of the backbone. In particular, the distribution modules aim at extracting the uncertainty from both ID and OOD data. We further provide theoretical analysis for the convergence of ABNN, and experimentally validate its superiority by comparing with some state-of-the-art uncertainty estimation methods Code will be made available.

cs.LG↗

arXiv2023

Bayesian Domain Invariant Learning via Posterior Generalization of Parameter Distributions

Domain invariant learning aims to learn models that extract invariant features over various training domains, resulting in better generalization to unseen target domains. Recently, Bayesian Neural Networks have achieved promising results in domain invariant learning, but most works concentrate on aligning features distributions rather than parameter distributions. Inspired by the principle of Bayesian Neural Network, we attempt to directly learn the domain invariant posterior distribution of network parameters. We first propose a theorem to show that the invariant posterior of parameters can be implicitly inferred by aggregating posteriors on different training domains. Our assumption is more relaxed and allows us to extract more domain invariant information. We also propose a simple yet effective method, named PosTerior Generalization (PTG), that can be used to estimate the invariant parameter distribution. PTG fully exploits variational inference to approximate parameter distributions, including the invariant posterior and the posteriors on training domains. Furthermore, we develop a lite version of PTG for widespread applications. PTG shows competitive performance on various domain generalization benchmarks on DomainBed. Additionally, PTG can use any existing domain generalization methods as its prior, and combined with previous state-of-the-art method the performance can be further improved. Code will be made public.

cs.LG↗