arXiv subjects

Xiang Zhu

Publications and source records attributed to Xiang Zhu.

At least 37 records · Page 2Linked to original sources

arXiv2025

Image Motion Blur Removal in the Temporal Dimension with Video Diffusion Models

Most motion deblurring algorithms rely on spatial-domain convolution models, which struggle with the complex, non-linear blur arising from camera shake and object motion. In contrast, we propose a novel single-image deblurring approach that treats motion blur as a temporal averaging phenomenon. Our core innovation lies in leveraging a pre-trained video diffusion transformer model to capture diverse motion dynamics within a latent space. It sidesteps explicit kernel estimation and effectively accommodates diverse motion patterns. We implement the algorithm within a diffusion-based inverse problem framework. Empirical results on synthetic and real-world datasets demonstrate that our method outperforms existing techniques in deblurring complex motion blur scenarios. This work paves the way for utilizing powerful video diffusion models to address single-image deblurring challenges.

eess.IV↗

arXiv2024

Advancing Humanoid Locomotion: Mastering Challenging Terrains with Denoising World Model Learning

Humanoid robots, with their human-like skeletal structure, are especially suited for tasks in human-centric environments. However, this structure is accompanied by additional challenges in locomotion controller design, especially in complex real-world environments. As a result, existing humanoid robots are limited to relatively simple terrains, either with model-based control or model-free reinforcement learning. In this work, we introduce Denoising World Model Learning (DWL), an end-to-end reinforcement learning framework for humanoid locomotion control, which demonstrates the world's first humanoid robot to master real-world challenging terrains such as snowy and inclined land in the wild, up and down stairs, and extremely uneven terrains. All scenarios run the same learned neural network with zero-shot sim-to-real transfer, indicating the superior robustness and generalization capability of the proposed method.

cs.RO↗

arXiv2024

RCDM: Enabling Robustness for Conditional Diffusion Model

The conditional diffusion model (CDM) enhances the standard diffusion model by providing more control, improving the quality and relevance of the outputs, and making the model adaptable to a wider range of complex tasks. However, inaccurate conditional inputs in the inverse process of CDM can easily lead to generating fixed errors in the neural network, which diminishes the adaptability of a well-trained model. The existing methods like data augmentation, adversarial training, robust optimization can improve the robustness, while they often face challenges such as high computational complexity, limited applicability to unknown perturbations, and increased training difficulty. In this paper, we propose a lightweight solution, the Robust Conditional Diffusion Model (RCDM), based on control theory to dynamically reduce the impact of noise and significantly enhance the model's robustness. RCDM leverages the collaborative interaction between two neural networks, along with optimal control strategies derived from control theory, to optimize the weights of two networks during the sampling process. Unlike conventional techniques, RCDM establishes a mathematical relationship between fixed errors and the weights of the two neural networks without incurring additional computational overhead. Extensive experiments were conducted on MNIST and CIFAR-10 datasets, and the results demonstrate the effectiveness and adaptability of our proposed model.

cs.LG↗

arXiv2024

Feature Re-Embedding: Towards Foundation Model-Level Performance in Computational Pathology

Multiple instance learning (MIL) is the most widely used framework in computational pathology, encompassing sub-typing, diagnosis, prognosis, and more. However, the existing MIL paradigm typically requires an offline instance feature extractor, such as a pre-trained ResNet or a foundation model. This approach lacks the capability for feature fine-tuning within the specific downstream tasks, limiting its adaptability and performance. To address this issue, we propose a Re-embedded Regional Transformer (R$^2$T) for re-embedding the instance features online, which captures fine-grained local features and establishes connections across different regions. Unlike existing works that focus on pre-training powerful feature extractor or designing sophisticated instance aggregator, R$^2$T is tailored to re-embed instance features online. It serves as a portable module that can seamlessly integrate into mainstream MIL models. Extensive experimental results on common computational pathology tasks validate that: 1) feature re-embedding improves the performance of MIL models based on ResNet-50 features to the level of foundation model features, and further enhances the performance of foundation model features; 2) the R$^2$T can introduce more significant performance improvements to various MIL models; 3) R$^2$T-MIL, as an R$^2$T-enhanced AB-MIL, outperforms other latest methods by a large margin.The code is available at: https://github.com/DearCaat/RRT-MIL.

cs.CV↗

arXiv2024

Multi-Objective Sizing Optimization Method of Microgrid Considering Cost and Carbon Emissions

Microgrid serves as a promising solution to integrate and manage distributed renewable energy resources. In this paper, we establish a stochastic multi-objective sizing optimization (SMOSO) model for microgrid planning, which fully captures the battery degradation characteristics and the total carbon emissions. The microgrid operator aims to simultaneously maximize the economic benefits and minimize carbon emissions, and the degradation of the battery energy storage system (BESS) is modeled as a nonlinear function of power throughput. A self-adaptive multi-objective genetic algorithm (SAMOGA) is proposed to solve the SMOSO model, and this algorithm is enhanced by pre-grouped hierarchical selection and self-adaptive probabilities of crossover and mutation. Several case studies are conducted to determine the microgrid size by analyzing Pareto frontiers, and the simulation results validate that the proposed method has superior performance over other algorithms on the solution quality of optimum and diversity.

eess.SY↗

arXiv2024

Weighted Voronoi-Delaunay dual on polyhedral surfaces and its finiteness

We aim to give a strict proof of the existence and uniqueness of the weighted Voronoi decomposition and the dual weighted Delaunay triangulation on Euclidean and hyperbolic polyhedral surface as well as hyperbolic surface with geodesic boundaries. Since the former definition of the Voronoi cell may not be simply connected, we slightly adjust the definition. Our proof is to construct an isotopic map instead of using the edge-flipping algorithm, which is a generalization of the one by Dyer et al. The main theorem of this paper is a lemma for proving the existence of the inversive distance circle packing.

math.DG↗

arXiv2024

A family of air-stable chalcogenide solid electrolytes in Li$_2$BMQ$_4$ (B = Ca, Sr and Ba; M = Si, Ge and Sn; Q = O, S and Se) systems

Combining high-throughput first-principles calculations and experimental measurements, we have identified a novel family of fast lithium-ion chalcogenide conductors in Li$_2$BMQ$_4$ (2114, B = Ca, Sr and Ba; M = Si, Ge and Sn; Q = O, S and Se) systems. Our calculations demonstrate that most of the thermodynamically and kinetically stable sulfides and selenides in this new system exhibit ultralow Li$^+$ ion migration activation energy (0.16 eV ~ 0.56 eV) and considerable bandgaps varying between ~ 2 eV and 3.5 eV. We have successfully synthesized Li$_2$BaSnS$_4$ and Li$_2$SrSiS$_4$, and they exhibit excellent moisture stability through H$_2$S gas measurements. Electrochemical impedance measurements indicate 2114 systems show the typical features of solid ionic conductors, with a room-temperature Li$^+$ conductivity close to 5$\times$10$^{-4}$ mS/cm aligning with our molecular dynamics simulations. Furthermore, we have theoretically investigated the substitution of Cl$^-$ at S$^{2-}$ site. The doped compounds display significantly higher conductivity, with an increase of about three orders of magnitude (up to a maximum of 0.72 mS/cm) compared to the undoped compounds. These findings offer valuable insights for the further exploration of potential chalcogenide solid electrolyte materials with robust air stability and enhanced ionic conductivity for practical applications in lithium-ion batteries.

cond-mat.mtrl-sci↗

arXiv2024

Noiseless linear amplification-based quantum Ziv-Zakai bound for phase estimation and its Heisenberg error limits in noisy scenarios

In this work, we address the central problem about how to effectively find the available precision limit of unknown parameters. In the framework of the quantum Ziv-Zakai bound (QZZB), we employ noiseless linear amplification (NLA)techniques to an initial coherent state (CS) as the probe state, and focus on whether the phase estimation performance is improved significantly in noisy scenarios, involving the photon-loss and phase-diffusion cases. More importantly, we also obtain two kinds of Heisenberg error limits of the QZZB with the NLA-based CS in these noisy scenarios, making comparisons with both the Margolus-Levitin (ML) type bound and the Mandelstam-Tamm (MT) type bound. Our analytical results show that in cases of photon loss and phase diffusion, the phase estimation performance of the QZZB can be improved remarkably by increasing the NLA gain factor. Particularly, the improvement is more pronounced with severe photon losses. Furthermore in minimal photon losses, our Heisenberg error limit shows better compactness than the cases of the ML-type and MT-type bounds. Our findings will provide an useful guidance for accomplishing more complex quantum information processing tasks.

quant-ph↗

arXiv2024

Roles of non-axisymmetric perturbations in free drift vertical displacement events on EAST

The safe operation of most tokamaks, especially the largen sized ones, rely on the feedback control of the vertical displacement events (VDEs). However, most these feedback control systems are based on the axisymmetric VDE models. In this work, we use NIMROD simulations to study the roles of non-axisymmetric perturbations in free drift vertical displacement events on EAST. The high-$n$ modes in non-axisymmetric VDE grow first, which drive the formation of high-$n$ magnetic island chains. Subsequently, the magnetic island chains grow and overlap with each other, leading to the destruction of the magnetic flux surface, which induces a minor disruption and accelerates the start of the following major disruption. The magnetic island and the stochastic magnetic field allow the toroidally asymmetric poloidal plasma current to jet towards the hoop force direction, forming the finger and filamentary structures. Such a plasma current asymmetry strongly depends on the anisotropy in thermal transport coefficients.

physics.plasm-ph↗

arXiv2023

The existence of inversive distance circle packing on hyperbolic polyhedral surface

In this paper, we prove that given a hyperbolic polyhedral metric with an inversive distance circle packing, and a target discrete curvature satisfying Gauss-Bonnet formula, there exist a unique inversive distance circle packing which is discrete conformal to the former one. We deform the surface by discrete Ricci flow, and do surgery by edge flipping when the orthogonal circles of some faces are about to be non-compact. The revised weighted Delaunay inequality of hyperbolic case implies the compactness of the orthogonal circle. We use a variational principle of a convex Ricci potential defined on the fiber bundles with cell-decomposition and differential structure based on Teichmüller space to finish the proof.

math.DG↗

arXiv2023

The existence of inversive distance circle packing on polyhedral surface

We prove that for any discrete curvature satisfying Gauss-Bonnet formula, there exist a unique up to scaling inversive distance circle packing in the discrete conformal equivalent class, whose polyhedral metric meets the target curvature. We prove it by constructing diffeomorphism between fiber bundles with cell decomposition based on Teichmüller spaces, and each discrete conformal equivalent class is a fiber passing through finite cell with respect to triangulations, which means we can do surgery on the discrete Ricci flow by edge flipping using a generalized Ptolemy equation to ensure it converge and never blow up.

math.DG↗

arXiv2023

Stylized Table Tennis Robots Skill Learning with Incomplete Human Demonstrations

In recent years, Reinforcement Learning (RL) is becoming a popular technique for training controllers for robots. However, for complex dynamic robot control tasks, RL-based method often produces controllers with unrealistic styles. In contrast, humans can learn well-stylized skills under supervisions. For example, people learn table tennis skills by imitating the motions of coaches. Such reference motions are often incomplete, e.g. without the presence of an actual ball. Inspired by this, we propose an RL-based algorithm to train a robot that can learn the playing style from such incomplete human demonstrations. We collect data through the teaching-and-dragging method. We also propose data augmentation techniques to enable our robot to adapt to balls of different velocities. We finally evaluate our policy in different simulators with varying dynamics.

cs.RO↗

arXiv2023

Global simulations of kinetic-magnetohydrodynamic processes with energetic electrons in tokamak plasmas

The energetic electrons (EEs) generated through auxiliary heating have been found to destabilize various Alfven eigenmodes (AEs) in recent experiments, which in turn lead to the EE transport and degrade the plasma energy confinement. In this work, we propose a global fluid-kinetic hybrid model for studying corresponding kinetic-magnetohydrodynamic (MHD) processes by coupling the drift-kinetic EEs to the Landau-fluid model of bulk plasmas in a non-perturbative manner. The numerical capability of Landau-fluid bulk plasmas is obtained based on a well-benchmarked eigenvalue code MAS [Multiscale Analysis of plasma Stabilities, J. Bao et al. Nucl. Fusion accepted 2023], and the EE responses to the electromagnetic fluctuations are analytically derived, which not only contribute to the MHD interchange drive and parallel current but also lead to the newly kinetic particle compression with the precessional drift resonance in the leading order. The hybrid model is casted into a nonlinear eigenvalue matrix equation and solved iteratively using Newton's method. By calibrating the EE precession frequency against the particle equation of motion in general geometry and applying more realistic trapped particle distribution in the poloidal plane, MAS simulations of EE-driven beta-induced Alfven eigenmodes (e-BAE) show excellent agreements with gyrokinetic particle-in-cell simulations, and the non-perturbative effects of EEs on e-BAE mode structure, growth rate and damping rate are demonstrated. With these efforts, the upgraded MAS greatly improves the computation efficiency for plasma problems related to deeply-trapped EEs, which is superior than initial-value simulations restricted by the stringent electron Courant condition regarding to the practical application of fast linear analysis.

physics.plasm-ph↗

arXiv2022

Incorporating Domain Knowledge through Task Augmentation for Front-End JavaScript Code Generation

Code generation aims to generate a code snippet automatically from natural language descriptions. Generally, the mainstream code generation methods rely on a large amount of paired training data, including both the natural language description and the code. However, in some domain-specific scenarios, building such a large paired corpus for code generation is difficult because there is no directly available pairing data, and a lot of effort is required to manually write the code descriptions to construct a high-quality training dataset. Due to the limited training data, the generation model cannot be well trained and is likely to be overfitting, making the model's performance unsatisfactory for real-world use. To this end, in this paper, we propose a task augmentation method that incorporates domain knowledge into code generation models through auxiliary tasks and a Subtoken-TranX model by extending the original TranX model to support subtoken-level code generation. To verify our proposed approach, we collect a real-world code generation dataset and conduct experiments on it. Our experimental results demonstrate that the subtoken-level TranX model outperforms the original TranX model and the Transformer model on our dataset, and the exact match accuracy of Subtoken-TranX improves significantly by 12.75% with the help of our task augmentation method. The model performance on several code categories has satisfied the requirements for application in industrial systems. Our proposed approach has been adopted by Alibaba's BizCook platform. To the best of our knowledge, this is the first domain code generation system adopted in industrial development environments.

cs.SE↗

arXiv2022

A Contact-Safe Reinforcement Learning Framework for Contact-Rich Robot Manipulation

Reinforcement learning shows great potential to solve complex contact-rich robot manipulation tasks. However, the safety of using RL in the real world is a crucial problem, since unexpected dangerous collisions might happen when the RL policy is imperfect during training or in unseen scenarios. In this paper, we propose a contact-safe reinforcement learning framework for contact-rich robot manipulation, which maintains safety in both the task space and joint space. When the RL policy causes unexpected collisions between the robot arm and the environment, our framework is able to immediately detect the collision and ensure the contact force to be small. Furthermore, the end-effector is enforced to perform contact-rich tasks compliantly, while keeping robust to external disturbances. We train the RL policy in simulation and transfer it to the real robot. Real world experiments on robot wiping tasks show that our method is able to keep the contact force small both in task space and joint space even when the policy is under unseen scenario with unexpected collision, while rejecting the disturbances on the main task.

cs.RO↗

arXiv2020

Super-Resolving Commercial Satellite Imagery Using Realistic Training Data

In machine learning based single image super-resolution, the degradation model is embedded in training data generation. However, most existing satellite image super-resolution methods use a simple down-sampling model with a fixed kernel to create training images. These methods work fine on synthetic data, but do not perform well on real satellite images. We propose a realistic training data generation model for commercial satellite imagery products, which includes not only the imaging process on satellites but also the post-process on the ground. We also propose a convolutional neural network optimized for satellite images. Experiments show that the proposed training data generation model is able to improve super-resolution performance on real satellite images.

cs.CV↗

arXiv2020

Combinatorial Calabi flow with surgery on surfaces

Computing uniformization maps for surfaces has been a challenging problem and has many practical applications. In this paper, we provide a theoretically rigorous algorithm to compute such maps via combinatorial Calabi flow for vertex scaling of polyhedral metrics on surfaces, which is an analogue of the combinatorial Yamabe flow introduced by Luo. To handle the singularies along the combinatorial Calabi flow, we do surgery on the flow by flipping. Using the discrete conformal theory established by Gu-Guo-Luo-Sun-Wu and Gu-Luo-Sun-Wu, we prove that for any initial Euclidean or hyperbolic polyhedral metric on a closed surface, the combinatorial Calabi flow with surgery exists for all time and converges exponentially fast after finite number of surgeries. The convergence is independent of the combinatorial structure of the initial triangulation on the surface.

math.GT↗

arXiv2005

Evolving Networks and Birth-and-Death Processes

Using a simple model with link removals as well as link additions, we show that an evolving network is scale free with a degree exponent in the range of (2, 4]. We then establish a relation between the network evolution and a set of non-homogeneous birth-and-death processes, and, with which, we capture the process by which the network connectivity evolves. We develop an effective algorithm to compute the network degree distribution accurately. Comparing analytical and numerical results with simulation, we identify some interesting network properties and verify the effectiveness of our method.

math-ph↗