arXiv subjects

Zhipeng Yu

Publications and source records attributed to Zhipeng Yu.

At least 19 recordsLinked to original sources

arXiv2026

The Price of Consistency: Exploiting Visual Anchors for Multimodal Jailbreaking in Video Generation

The rapid evolution of video generation has shifted the paradigm from pure text-driven to multi-conditional controllable generation, with reference images now widely adopted as conditional inputs to achieve superior spatiotemporal consistency. While these reference images serve as powerful visual anchors that significantly enhance controllability, their impact on safety remains largely unexplored. In this work, we reveal the visual anchoring effect: by enforcing consistency, the mechanism prevents the generated content from drifting away from the original harmful intent, thereby eliminating the model's natural safety escape route from harmful to benign content. Consequently, visual anchors inherently increase the safety risk---this is the price of consistency. Building on this insight, we propose Decoupling Intent via Visual Anchors (DIVA), a training-free multimodal jailbreak framework for video generation that exploits this vulnerability. DIVA decouples harmful intent into a static visual anchor image and a dynamic motion text prompt, and employs dual-criteria selection to balance attack stealthiness with semantic preservation. Extensive experiments across various leading commercial platforms and mainstream open-source video generation models demonstrate that DIVA achieves a substantially higher Attack Success Rate than existing text-only methods. To facilitate future research, we additionally contribute TI2VSafetyBench, the first safety benchmark for multi-conditional video generation.

cs.CV↗

arXiv2026

Correlation invariance unlocks robust calibration-free orbital-angular-momentum multiplexing transmission under dynamic scattering scenarios

Orbital angular momentum (OAM) multiplexing offers a promising approach to high-capacity optical communication by harnessing the orthogonality of vortex beams. However, its practical deployment is severely limited in real-world settings where dynamic scattering media, such as turbulent atmosphere, distort multiplexed fields into random speckles and disrupt OAM demultiplexing. Although existing wavefront shaping and deep learning methods can mitigate static distortions, they fail under time-varying scattering conditions, leading to significant crosstalk and unreliable recovery. Here, we introduce a new concept, correlation invariance, which enables scattering-immune, robust OAM multiplexed transmission through dynamic media. By capturing orthogonally polarized speckle holograms in a compact common-path geometry and computing their intensity cross-correlation, dynamically imposed scattering phases are cancelled out while deterministic object information is preserved. This allows single-shot reconstruction of both amplitude and phase of the input OAM-multiplexed fields, without any pre-calibration or training. As a proof of principle, we demonstrate high-fidelity transmission of 24-bit RGB data with 99.61% accuracy under static scattering and 98.97% accuracy under dynamic scattering. This approach addresses a long-standing barrier in OAM-based systems and opens avenues for robust high-capacity optical communications, encryption, and imaging in dynamic scattering environments.

physics.optics↗

arXiv2026

Real-time Calibration-free Imaging Through Dynamic and Distinct Multimode Fibers via Spatial Harmonic Invariant Nonlinear Encoding (SHINE)

Multimode fibers (MMFs) provide a compact, high-throughput platform for minimally invasive imaging and information transmission. However, their utility is fundamentally constrained by mode mixing, which renders image transmission spatially disrupted and sensitive to external perturbations. Current imaging methods typically rely on transmission matrix measurement or deep learning models that are fragile to fiber movement, necessitating frequent, time-consuming calibrations and re-calibrations that are easily disrupted and fail to generalize across different fiber configurations, let alone across entirely distinct fibers. Here, we propose a calibration and feedback-free MMF coherent imaging paradigm, that we termed Spatial Harmonic Invariant Nonlinear Encoding (SHINE). By leveraging the angle-dependent phase-matching conditions of second-harmonic generation, we encode spatial features into broadband spectral signatures that possess intrinsic insensitivity not only to modal scrambling but also to fiber bending, movement, as well as structural variations. This spectral representation enables a deep learning model to robustly reconstruct images in real time despite dynamic perturbations and even generalizes well to distinct MMFs without recalibration or feedback. We achieve experimentally an average Pearson correlation coefficient (PCC) of 0.82 for image reconstruction tasks on Fashion-MNIST and a classification accuracy of 92.3% on HERLEV biomedical dataset. Uniquely, our method exhibits remarkable cross-fiber generalization: a model trained on a single MMF successfully reconstructs images transmitted through entirely distinct, previously unseen MMFs with a PCC of 0.74. These results establish a robust, calibration-free framework for imaging through MMFs in real time, paving the way for practical, resilient optical diagnostics that operate without distal-end feedback.

physics.optics↗

arXiv2025

Falcon: A Comprehensive Chinese Text-to-SQL Benchmark for Enterprise-Grade Evaluation

We introduce Falcon, a cross-domain Chinese text-to-SQL benchmark grounded in an enterprise-compatible dialect (MaxCompute/Hive). It contains 600 Chinese questions over 28 databases; 77% require multi-table reasoning and over half touch more than four tables. Each example is annotated along SQL-computation features and Chinese semantics. For evaluation, we release a robust execution comparator and an automated evaluation pipeline, under which all current state-of-the-art large-scale models (including Deepseek) achieve accuracies of at most 50%. Major errors originate from two sources: (1) schema linking in large enterprise landscapes - hundreds of tables, denormalized fields, ambiguous column names, implicit foreign-key relations and domain-specific synonyms that make correct join/column selection difficult; and (2) mapping concise, colloquial Chinese into the exact operators and predicates required for analytics - e.g., choosing the correct aggregation and group-by keys, expressing time windows and granularities, applying unit conversions, handling NULLs and data-quality rules, and formulating nested or windowed subqueries. Falcon therefore targets Chinese-specific semantics and enterprise dialects (abbreviations, business jargon, fuzzy entity references) and provides a reproducible middle ground before full production deployment by using realistic enterprise schemas, query templates, an execution comparator, and an automated evaluation pipeline for end-to-end validation.

cs.CL↗

arXiv2025

Implementing van der Waals forces for polytope particles in DEM simulations of clay

Clay minerals are non-spherical nano-scale particles that usually form flocculated, house-of-card like structures under the influence of inter-molecular forces. Numerical modeling of clays is still in its infancy as the required inter-particle forces are available only for spherical particles. A polytope approach would allow shape-accurate forces and torques while simultaneously being more performant. The Anandarajah solution provides an analytical formulation for van der Waals forces for cuboid particles but in its original form is not suitable for implementation in DEM simulations. In this work, we discuss the necessary changes for a functional implementation of the Anandarajah solution in a DEM simulation of rectangular particles and their extension to cuboid particles.

cond-mat.soft↗

arXiv2025

HoloPart: Generative 3D Part Amodal Segmentation

3D part amodal segmentation--decomposing a 3D shape into complete, semantically meaningful parts, even when occluded--is a challenging but crucial task for 3D content creation and understanding. Existing 3D part segmentation methods only identify visible surface patches, limiting their utility. Inspired by 2D amodal segmentation, we introduce this novel task to the 3D domain and propose a practical, two-stage approach, addressing the key challenges of inferring occluded 3D geometry, maintaining global shape consistency, and handling diverse shapes with limited training data. First, we leverage existing 3D part segmentation to obtain initial, incomplete part segments. Second, we introduce HoloPart, a novel diffusion-based model, to complete these segments into full 3D parts. HoloPart utilizes a specialized architecture with local attention to capture fine-grained part geometry and global shape context attention to ensure overall shape consistency. We introduce new benchmarks based on the ABO and PartObjaverse-Tiny datasets and demonstrate that HoloPart significantly outperforms state-of-the-art shape completion methods. By incorporating HoloPart with existing segmentation techniques, we achieve promising results on 3D part amodal segmentation, opening new avenues for applications in geometry editing, animation, and material assignment.

cs.CV↗

arXiv2025

TripoSG: High-Fidelity 3D Shape Synthesis using Large-Scale Rectified Flow Models

Recent advancements in diffusion techniques have propelled image and video generation to unprecedented levels of quality, significantly accelerating the deployment and application of generative AI. However, 3D shape generation technology has so far lagged behind, constrained by limitations in 3D data scale, complexity of 3D data processing, and insufficient exploration of advanced techniques in the 3D domain. Current approaches to 3D shape generation face substantial challenges in terms of output quality, generalization capability, and alignment with input conditions. We present TripoSG, a new streamlined shape diffusion paradigm capable of generating high-fidelity 3D meshes with precise correspondence to input images. Specifically, we propose: 1) A large-scale rectified flow transformer for 3D shape generation, achieving state-of-the-art fidelity through training on extensive, high-quality data. 2) A hybrid supervised training strategy combining SDF, normal, and eikonal losses for 3D VAE, achieving high-quality 3D reconstruction performance. 3) A data processing pipeline to generate 2 million high-quality 3D samples, highlighting the crucial rules for data quality and quantity in training 3D generative models. Through comprehensive experiments, we have validated the effectiveness of each component in our new framework. The seamless integration of these parts has enabled TripoSG to achieve state-of-the-art performance in 3D shape generation. The resulting 3D shapes exhibit enhanced detail due to high-resolution capabilities and demonstrate exceptional fidelity to input images. Moreover, TripoSG demonstrates improved versatility in generating 3D models from diverse image styles and contents, showcasing strong generalization capabilities. To foster progress and innovation in the field of 3D generation, we will make our model publicly available.

cs.CV↗

arXiv2025

Enhancing Sample Utilization in Noise-Robust Deep Metric Learning With Subgroup-Based Positive-Pair Selection

The existence of noisy labels in real-world data negatively impacts the performance of deep learning models. Although much research effort has been devoted to improving the robustness towards noisy labels in classification tasks, the problem of noisy labels in deep metric learning (DML) remains under-explored. Existing noisy label learning methods designed for DML mainly discard suspicious noisy samples, resulting in a waste of the training data. To address this issue, we propose a noise-robust DML framework with SubGroup-based Positive-pair Selection (SGPS), which constructs reliable positive pairs for noisy samples to enhance the sample utilization. Specifically, SGPS first effectively identifies clean and noisy samples by a probability-based clean sample selectionstrategy. To further utilize the remaining noisy samples, we discover their potential similar samples based on the subgroup information given by a subgroup generation module and then aggregate them into informative positive prototypes for each noisy sample via a positive prototype generation module. Afterward, a new contrastive loss is tailored for the noisy samples with their selected positive pairs. SGPS can be easily integrated into the training process of existing pair-wise DML tasks, like image retrieval and face recognition. Extensive experiments on multiple synthetic and real-world large-scale label noise datasets demonstrate the effectiveness of our proposed method. Without any bells and whistles, our SGPS framework outperforms the state-of-the-art noisy label DML methods. Code is available at \url{https://github.com/smuelpeng/SGPS-NoiseFreeDML}.

cs.CV↗

arXiv2024

Groups of matrices with approximately submultiplicative spectra

We say that a semigroup of matrices has a submultiplicative spectrum if the spectrum of the product of any two elements of the semigroup is contained in the product of the two spectra in question (as sets). In this note we explore an approximate version of this condition.

math.RT↗

arXiv2024

Exploring the Optimal Size of Grid-forming Energy Storage in an Off-grid Renewable P2H System under Multi-timescale Energy Management

Utility-scale off-grid renewable power-to-hydrogen systems (OReP2HS), typically comprising photovoltaic plants, wind turbines, electrolyzers (ELs), and battery energy storage system (BESS), requires at least one component, generally the BESS, working with grid-forming ability to provide frequency and voltage references and regulate them through transient power support. However, existing designs of OReP2HS based on the energy management strategies (EMSs) with 5-min or hourly resolution fail to capture fast transients and may underestimate the BESS size required to ensure adequate grid-forming ability. This paper first proposes a framework of multi-timescale EMS covers from those for power system transient behaviors to second-level EL load adjustments and minute-level intra-day scheduling to coordinate renewable power, BESS, and ELs. Then, an iterative search procedure based on high-fidelity simulation is employed to determine the cost-effective BESS size that satisfies grid-forming, long-term energy balancing over 8760 hours, and emergency support requirements. Case studies based on a planned OReP2HS project in Inner Mongolia, China, show that the proposed EMS yields a base-case LCOH of 33.212 CNY/kg, with CAPEX of BESS accounting for 17.83% of total investment. The optimal BESS capacity represents 13.6% of the rated hourly renewable output and shows a yearly degradation of 4.87%. Sensitivity analysis reveals that reducing the electrolytic load adjustment time step from 90 to 5 seconds and increasing its ramping limit from 1% to 10% rated power per second, the BESS size decreases by 53.57%, and the LCOH decreases to 25.458 CNY/kg. Considering the cost of designing and manufacturing utility-scale ELs with fast load regulation capability, a load adjustment time step of 5 to 10 seconds and a ramping limit of 4-6% rated power per second are recommended.

math.OC↗

arXiv2024

Estimating Long-term Heterogeneous Dose-response Curve: Generalization Bound Leveraging Optimal Transport Weights

Long-term treatment effect estimation is a significant but challenging problem in many applications. Existing methods rely on ideal assumptions, such as no unobserved confounders or binary treatment, to estimate long-term average treatment effects. However, in numerous real-world applications, these assumptions could be violated, and average treatment effects are insufficient for personalized decision-making. In this paper, we address a more general problem of estimating long-term Heterogeneous Dose-Response Curve (HDRC) while accounting for unobserved confounders and continuous treatment. Specifically, to remove the unobserved confounders in the long-term observational data, we introduce an optimal transport weighting framework to align the long-term observational data to an auxiliary short-term experimental data. Furthermore, to accurately predict the heterogeneous effects of continuous treatment, we establish a generalization bound on counterfactual prediction error by leveraging the reweighted distribution induced by optimal transport. Finally, we develop a long-term HDRC estimator building upon the above theoretical foundations. Extensive experiments on synthetic and semi-synthetic datasets demonstrate the effectiveness of our approach.

cs.LG↗

arXiv2023

Triplane Meets Gaussian Splatting: Fast and Generalizable Single-View 3D Reconstruction with Transformers

Recent advancements in 3D reconstruction from single images have been driven by the evolution of generative models. Prominent among these are methods based on Score Distillation Sampling (SDS) and the adaptation of diffusion models in the 3D domain. Despite their progress, these techniques often face limitations due to slow optimization or rendering processes, leading to extensive training and optimization times. In this paper, we introduce a novel approach for single-view reconstruction that efficiently generates a 3D model from a single image via feed-forward inference. Our method utilizes two transformer-based networks, namely a point decoder and a triplane decoder, to reconstruct 3D objects using a hybrid Triplane-Gaussian intermediate representation. This hybrid representation strikes a balance, achieving a faster rendering speed compared to implicit representations while simultaneously delivering superior rendering quality than explicit representations. The point decoder is designed for generating point clouds from single images, offering an explicit representation which is then utilized by the triplane decoder to query Gaussian features for each point. This design choice addresses the challenges associated with directly regressing explicit 3D Gaussian attributes characterized by their non-structural nature. Subsequently, the 3D Gaussians are decoded by an MLP to enable rapid rendering through splatting. Both decoders are built upon a scalable, transformer-based architecture and have been efficiently trained on large-scale 3D datasets. The evaluations conducted on both synthetic datasets and real-world images demonstrate that our method not only achieves higher quality but also ensures a faster runtime in comparison to previous state-of-the-art techniques. Please see our project page at https://zouzx.github.io/TriplaneGaussian/.

cs.CV↗

arXiv2023

Speckle-Driven Single-Shot Orbital Angular Momentum Recognition with Ultra-Low Sampling Density

Orbital angular momentum (OAM) recognition of vortex beams is critical for applications ranging from optical communications to quantum technologies. However, conventional approaches designed for free-space propagation struggle when light passes through scattering media, such as multimode fibers (MMF), and often rely on high-resolution sensors with tens of thousands of pixels to record detailed intensity profiles. Here, by harnessing scattering media as intrinsic encoders rather than detrimental factors, we introduce a speckle-driven OAM recognition technique termed patially multiplexed points detection (SMPD). This method extracts intensity information from a few spatially distributed points in a speckle plane, where object feature is naturally multiplexed, thereby transforming scattering from a detrimental effect into an efficient encoding mechanism while drastically reducing sampling requirements. Remarkably, it achieves over 99% retrieval accuracy for OAMs recognition using just 16 sampling points, corresponding to a sampling density of 0.024% compared with conventional imaging-based approaches. Furthermore, spatiotemporally interleaved vortex beams decoding, highcapacity OAM-multiplexed communication, MNIST, and Fashion-MNIST classification are implemented to verify the versatility of SMPD. This work establishes a scalable strategy for efficient optical information processing and fiberbased sensing in complex environments.

physics.optics↗

arXiv2023

Multi-Stage Expansion Planning for Decarbonizing Thermal Generation Supported Renewable Power Systems Using Hydrogen and Ammonia Storage

Large-scale centralized development of wind and solar energy and peer-to-grid transmission of renewable energy source (RES) via high voltage direct current (HVDC) has been regarded as one of the most promising ways to achieve goals of peak carbon and carbon neutrality in China. Traditionally, large-scale thermal generation is needed to economically support the load demand of HVDC with a given profile, which in turn raises concerns about carbon emissions. To address the issues above, hydrogen energy storage system (HESS) and ammonia energy storage system (AESS) are introduced to gradually replace thermal generation, which is represented as a multi-stage expansion planning (MSEP) problem. Specifically, first, HESS and AESS are established in the MSEP model with carbon emission reduction constraints, and yearly data with hourly time resolution are utilized for each stage to well describe the intermittence of RES. Then, a combined Dantzig-Wolfe decomposition (DWD) and column generation (CG) solution approach is proposed to efficiently solve the large-scale MSEP model. Finally, a real-life system in China is studied. The results indicate that HESS and AESS have the potential to handle the intermittence of RES, as well as the monthly imbalance between RES and load demand. Especially under the goal of carbon neutrality, the contribution of HESS and AESS in reducing levelized cost of energy (LCOE) reaches 12.28% and 14.59%, respectively, which finally leads to a LCOE of 0.4324 RMB/kWh.

eess.SY↗

arXiv2023

Towards Prompt-robust Face Privacy Protection via Adversarial Decoupling Augmentation Framework

Denoising diffusion models have shown remarkable potential in various generation tasks. The open-source large-scale text-to-image model, Stable Diffusion, becomes prevalent as it can generate realistic artistic or facial images with personalization through fine-tuning on a limited number of new samples. However, this has raised privacy concerns as adversaries can acquire facial images online and fine-tune text-to-image models for malicious editing, leading to baseless scandals, defamation, and disruption to victims' lives. Prior research efforts have focused on deriving adversarial loss from conventional training processes for facial privacy protection through adversarial perturbations. However, existing algorithms face two issues: 1) they neglect the image-text fusion module, which is the vital module of text-to-image diffusion models, and 2) their defensive performance is unstable against different attacker prompts. In this paper, we propose the Adversarial Decoupling Augmentation Framework (ADAF), addressing these issues by targeting the image-text fusion module to enhance the defensive performance of facial privacy protection algorithms. ADAF introduces multi-level text-related augmentations for defense stability against various attacker prompts. Concretely, considering the vision, text, and common unit space, we propose Vision-Adversarial Loss, Prompt-Robust Augmentation, and Attention-Decoupling Loss. Extensive experiments on CelebA-HQ and VGGFace2 demonstrate ADAF's promising performance, surpassing existing algorithms.

cs.CV↗

arXiv2023

Optimal Sizing of Isolated Renewable Power Systems with Ammonia Synthesis: Model and Solution Approach

Isolated renewable power to ammonia (IRePtA) has been recognized as a promising way to decarbonize the chemical industry. Optimal sizing of the renewable power system is significant to improve the techno-economic of IRePtA since the investment of power sources exceeds 80\% of the total investment. However, multi-timescale electricity, hydrogen, and ammonia storages, minimum power supply for system safety, and the multi-year uncertainty of renewable generation lead to difficulties in planning. To address the issues above, an IGDT-MILFP model is proposed. First, the levelized cost of ammonia (LCOA) is directly formulated as the objective, rendering a mixed integer linear fractional programming (MILFP) problem. Information gap decision theory (IGDT) is utilized to handle the multi-year uncertainty of renewable generation. Second, a combined Charnes-Cooper (C&C) transformation and Branch-and-Bound (B&B) method is proposed to efficiently solve the large-scale IGDT-MILFP model, giving robust and opportunistic planning results. Then, Markov Chain Monte Carlo (MCMC) sampling-based posteriori analysis is leveraged to quantify the long-run performance. Finally, a real-life system in Inner Mongolia, China, is studied. The results indicate that the proposed methods could reduce the computational burden by orders of magnitude for solving a large-scale MILFP problem. Moreover, the proposed IGDT-MILFP model is necessary and accurate to obtain an optimal capacity allocation with the lowest expected LCOA (3610 RMB/t) in long-run simulations.

eess.SY↗

arXiv2023

Multi-timescale Trading Strategy for Renewable Power to Ammonia Virtual Power Plant in the Electricity, Hydrogen, and Ammonia Markets

Renewable power to ammonia (RePtA) is a prominent zero-carbon pathway for decarbonization. Due to the imbalance between renewables and production energy demand, the RePtA system relies on the electricity exchange with the power grid. Participating in the electricity market as a virtual power plant (VPP) may help to reduce energy costs. However, the power profile of local photovoltaics and wind turbines is similar to those in the market, resulting in rising energy costs under the conventional strategy. Hence, we develop a multi-timescale trading strategy for the RePtA VPP in the electricity, hydrogen, and ammonia markets. By utilizing the hydrogen and ammonia buffer systems, the RePtA VPP can optimally coordinate production planning. Moreover, we find it possible to describe the trading of electricity, ammonia, and hydrogen in a unified framework. The two-stage robust optimization model of the electricity market is extended to multiple markets and solved by the column and constraint generation (CC\&G) algorithm. The case is derived from an actual project in the Inner Mongolia Autonomous Region. Sensitivity analysis demonstrates the economic advantages of an RePtA VPP joining multiple markets over conventional strategy and reveals the necessity of the hydrogen and ammonia buffer and reactor's flexibility.

eess.SY↗

arXiv2023

HMDO: Markerless Multi-view Hand Manipulation Capture with Deformable Objects

We construct the first markerless deformable interaction dataset recording interactive motions of the hands and deformable objects, called HMDO (Hand Manipulation with Deformable Objects). With our built multi-view capture system, it captures the deformable interactions with multiple perspectives, various object shapes, and diverse interactive forms. Our motivation is the current lack of hand and deformable object interaction datasets, as 3D hand and deformable object reconstruction is challenging. Mainly due to mutual occlusion, the interaction area is difficult to observe, the visual features between the hand and the object are entangled, and the reconstruction of the interaction area deformation is difficult. To tackle this challenge, we propose a method to annotate our captured data. Our key idea is to collaborate with estimated hand features to guide the object global pose estimation, and then optimize the deformation process of the object by analyzing the relationship between the hand and the object. Through comprehensive evaluation, the proposed method can reconstruct interactive motions of hands and deformable objects with high quality. HMDO currently consists of 21600 frames over 12 sequences. In the future, this dataset could boost the research of learning-based reconstruction of deformable interaction scenes.

cs.CV↗