arXiv subjects

Yang Jiao

Publications and source records attributed to Yang Jiao.

At least 55 records · Page 3Linked to original sources

arXiv2025

Cellular Traffic Prediction via Byzantine-robust Asynchronous Federated Learning

Network traffic prediction plays a crucial role in intelligent network operation. Traditional prediction methods often rely on centralized training, necessitating the transfer of vast amounts of traffic data to a central server. This approach can lead to latency and privacy concerns. To address these issues, federated learning integrated with differential privacy has emerged as a solution to improve data privacy and model robustness in distributed settings. Nonetheless, existing federated learning protocols are vulnerable to Byzantine attacks, which may significantly compromise model robustness. Developing a robust and privacy-preserving prediction model in the presence of Byzantine clients remains a significant challenge. To this end, we propose an asynchronous differential federated learning framework based on distributionally robust optimization. The proposed framework utilizes multiple clients to train the prediction model collaboratively with local differential privacy. In addition, regularization techniques have been employed to further improve the Byzantine robustness of the models. We have conducted extensive experiments on three real-world datasets, and the results elucidate that our proposed distributed algorithm can achieve superior performance over existing methods.

cs.LG↗

arXiv2025

OmniGenBench: A Benchmark for Omnipotent Multimodal Generation across 50+ Tasks

Recent breakthroughs in large multimodal models (LMMs), such as the impressive GPT-4o-Native, have demonstrated remarkable proficiency in following general-purpose instructions for image generation. However, current benchmarks often lack the necessary breadth and depth to fully evaluate the diverse capabilities of these models. To overcome this limitation, we introduce OmniGenBench, a novel and comprehensive benchmark meticulously designed to assess the instruction-following abilities of state-of-the-art LMMs across both perception-centric and cognition-centric dimensions. Our OmniGenBench includes 57 diverse sub-tasks grounded in real-world scenarios, systematically categorized according to the specific model capabilities they demand. For rigorous evaluation, we further employ a dual-mode protocol. This protocol utilizes off-the-shelf visual parsing tools for perception-centric tasks and a powerful LLM-based judger for cognition-centric tasks to assess the alignment between generated images and user instructions. Using OmniGenBench, we evaluate mainstream generative models, including prevalent models like GPT-4o, Gemini-2.0-Flash, and Seedream, and provide in-depth comparisons and analyses of their performance.Code and data are available at https://github.com/emilia113/OmniGenBench.

cs.CV↗

arXiv2025

Look Before You Decide: Prompting Active Deduction of MLLMs for Assumptive Reasoning

Recently, Multimodal Large Language Models (MLLMs) have achieved significant success across multiple disciplines due to their exceptional instruction-following capabilities and extensive world knowledge. However, whether these MLLMs possess human-like compositional reasoning abilities remains an open problem. To unveil their reasoning behaviors, we first curate a \textbf{M}ultimodal \textbf{A}ssumptive \textbf{R}ea\textbf{s}oning Benchmark (MARS-Bench) in this paper. Interestingly, we find that most prevalent MLLMs can be easily fooled by the introduction of a presupposition into the question, whereas such presuppositions appear naive to human reasoning. Besides, we also propose a simple yet effective method, Active Deduction (AD), a novel reinforcement learning paradigm to encourage the model to actively perform composite deduction before reaching a final decision. Equipped with the proposed AD method, a MLLM demonstrates significant improvements in assumptive reasoning abilities without compromising its general-purpose question-answering performance. We also provide extensive evaluations of both open-source and private MLLMs on MARS-Bench, along with experimental analyses of the AD method.

cs.CV↗

arXiv2025

Structure-Property Relationship in Disordered Hyperuniform Materials: Microstructure Representation, Field Fluctuations and Effective Properties

Disordered hyperuniform (DHU) materials are an emerging class of exotic heterogeneous material systems characterized by a unique combination of disordered local structures and a hidden long-range order, which endow them with unusual physical properties. Here, we consider material systems possessing continuously varying local material properties $\mathcal{K}({\bf x})$ modeled via a random field. We devise quantitative microstructure representation of the material systems based on a class of analytical spectral density function ${\tilde χ}_{_\mathcal{K}}({k})$ associated with $\mathcal{K}({\bf x})$, possessing a power-law small-$k$ scaling behavior ${\tilde χ}_{_\mathcal{K}}({k}) \sim k^α$. By controlling the exponent $α$ and using a highly efficient forward generative model, we obtain realizations of a wide spectrum of distinct material microstructures spanning from hyperuniform ($α>0$) to nonhyperuniform ($α=0$) to antihyperuniform ($α<0$) systems. We perform a comprehensive perturbation analysis to quantitatively connect the fluctuations of the local material property to the fluctuations of the resulting physical fields. In the weak-contrast limit, our first-order perturbation theory reveals that the physical fields associated with Class-I hyperuniform materials (characterized by $α\ge 2$) are also hyperuniform, albeit with a lower hyperuniformity exponent ($α-2$). As one moves away from this weak-contrast limit, the fluctuations of the physical field develop a diverging spectral density at the origin. We also establish an end-to-end mapping connecting the spectral density of the local material property to the overall effective conductivity of the material system via numerical homogenization.

cond-mat.mtrl-sci↗

arXiv2025

UniToken: Harmonizing Multimodal Understanding and Generation through Unified Visual Encoding

We introduce UniToken, an auto-regressive generation model that encodes visual inputs through a combination of discrete and continuous representations, enabling seamless integration of unified visual understanding and image generation tasks. Unlike previous approaches that rely on unilateral visual representations, our unified visual encoding framework captures both high-level semantics and low-level details, delivering multidimensional information that empowers heterogeneous tasks to selectively assimilate domain-specific knowledge based on their inherent characteristics. Through in-depth experiments, we uncover key principles for developing a unified model capable of both visual understanding and image generation. Extensive evaluations across a diverse range of prominent benchmarks demonstrate that UniToken achieves state-of-the-art performance, surpassing existing approaches. These results establish UniToken as a robust foundation for future research in this domain. The code and models are available at https://github.com/SxJyJay/UniToken.

cs.CV↗

arXiv2025

From point patterns to networks: to what extent does the Delaunay triangulation reproduce key spatial and density information?

It is important that a spatial network's construction algorithm reproduces the structural properties of the original physical embedding. Here, we assess the Delaunay triangulation as a spatial network construction algorithm for seven different types of 2D point patterns, including hyperuniform systems. The latter are characterized by completely suppressed normalized infinite-wavelength density fluctuations. We demonstrate that the quartile coefficients of dispersion of multiple centrality measures are capable of rank-ordering hyperuniform and nonhyperuniform systems independently, but they cannot distinguish a system that is nearly hyperuniform from hyperuniform systems. Thus, we investigate the local densities of the point pattern and of the network. We reveal that there is a strong correlation between local densities in the point pattern and network in nonhyperuniform systems, but there is no such correlation in hyperuniform systems. When calculating the pair-correlation function and local density covariance function on the point pattern and network, the point pattern and network functions are similar only in nonhyperuniform systems. In hyperuniform systems, the triangulation has a positive covariance of local network densities in pairs of nodes that are close together that is not present in the point patterns. Thus, we demonstrate that the Delaunay triangulation accurately captures the density fluctuations of the point pattern only when the point pattern possesses a positive local density covariance at small distances. Such positive correlation is seen in most real-world systems, so the Delaunay triangulation is generally an effective tool for building a spatial network from a 2D point pattern, but there are situations (i.e., disordered hyperuniform systems) where we caution that the Delaunay triangulation would not be effective at capturing the underlying physical embedding.

cond-mat.stat-mech↗

arXiv2025

Revealing Microscopic Objects in Fluorescence Live Imaging by Video-to-video Translation Based on A Spatial-temporal Generative Adversarial Network

In spite of being a valuable tool to simultaneously visualize multiple types of subcellular structures using spectrally distinct fluorescent labels, a standard fluoresce microscope is only able to identify a few microscopic objects; such a limit is largely imposed by the number of fluorescent labels available to the sample. In order to simultaneously visualize more objects, in this paper, we propose to use video-to-video translation that mimics the development process of microscopic objects. In essence, we use a microscopy video-to-video translation framework namely Spatial-temporal Generative Adversarial Network (STGAN) to reveal the spatial and temporal relationships between the microscopic objects, after which a microscopy video of one object can be translated to another object in a different domain. The experimental results confirm that the proposed STGAN is effective in microscopy video-to-video translation that mitigates the spectral conflicts caused by the limited fluorescent labels, allowing multiple microscopic objects be simultaneously visualized.

eess.IV↗

arXiv2025

Anomalous suppression of large-scale density fluctuations in classical and quantum spin liquids

Classical spin liquids (CSLs) are intriguing states of matter that do not exhibit long-range magnetic order and are characterized by an extensive ground-state degeneracy. Adding quantum fluctuations, which induce dynamics between these different classical ground states, can give rise to quantum spin liquids (QSLs). QSLs are highly entangled quantum phases of matter characterized by fascinating emergent properties, such as fractionalized excitations and topological order. One such exotic quantum liquid is the $\mathbb{Z}_2$ QSL, which can be regarded as a resonating valence bond (RVB) state formed from superpositions of dimer coverings of an underlying lattice. In this work, we unveil a \textit{hidden} large-scale structural property of archetypal CSLs and QSLs known as hyperuniformity, i.e., normalized infinite-wavelength density fluctuations are completely suppressed in these systems. In particular, we first demonstrate that classical ensembles of close-packed dimers and their corresponding quantum RVB states are perfectly hyperuniform in general. Subsequently, we focus on a ruby-lattice spin liquid that was recently realized in a Rydberg-atom quantum simulator, and show that the QSL remains effectively hyperuniform even in the presence of a finite density of spinon and vison excitations, as long as the dimer constraint is still largely preserved. Moreover, we demonstrate that metrics based on the framework of hyperuniformity can be used to distinguish the QSL from other proximate quantum phases. These metrics can help identify potential QSL candidates, which can then be further analyzed using more advanced, computationally-intensive quantum numerics to confirm their status as true QSLs.

cond-mat.str-el↗

arXiv2025

Understanding the Practice, Perception, and Challenge of Blind or Low Vision Students Learning through Accessible Technologies in Non-Inclusive 'Blind Colleges'

In developing and underdeveloped regions, many 'Blind Colleges' exclusively enroll individuals with Blindness or Vision Impairment (BLV) for higher education. While advancements in accessible technologies have facilitated BLV student integration into 'Integrated Colleges,' their implementation in 'Blind Colleges' remains uneven due to complex economic, social, and policy challenges. This study investigates the practices, perceptions, and challenges of BLV students using accessible technologies in a Chinese 'Blind College' through a two-part empirical approach. Our findings demonstrate that tactile and digital technologies enhance access to education but face significant integration barriers. We emphasize the critical role of early education in addressing capability gaps, BLV students' aspirations for more inclusive educational environments, and the systemic obstacles within existing frameworks. We advocate for leveraging accessible technologies to transition 'Blind Colleges' into 'Integrated Colleges,' offering actionable insights for policymakers, designers, and educators. Finally, we outline future research directions on accessible technology innovation and its implications for BLV education in resource-constrained settings.

cs.HC↗

arXiv2025

A dimension reduction procedure for the design of lattice-spring systems with minimal fabrication cost and required multi-functional properties

We show that the problem of the design of the lattices of elastoplastic current conducting springs with optimal multi-functional properties leads to an analytically tractable problem. Specifically, focusing on a lattice with a small number of springs, we use the technique of inequalities to reduce the number variables and to compute the minimal cost of lattice fabrication explicitly.

math.OC↗

arXiv2024

Three-Dimensional Construction of Hyperuniform, Nonhyperuniform and Antihyperuniform Random Media via Spectral Density Functions and Their Transport Properties

Rigorous theories connecting physical properties of a heterogeneous material to its microstructure offer a promising avenue to guide the computational material design and optimization. We present here an efficient Fourier-space based computational framework and employ a variety of analytical ${\tilde χ}_{_V}({k})$ functions that satisfy all known necessary conditions to construct 3D disordered stealthy hyperuniform, standard hyperuniform, nonhyperuniform, and antihyperuniform two-phase heterogeneous material systems at varying phase volume fractions. We show that a rich spectrum of distinct structures within each of the above classes of materials can be generated by tuning correlations in the system across length scales. We present the first realization of antihyperuniform two-phase heterogeneous materials in 3D, which are characterized by a power-law autocovariance function $χ_{_V}(r)$ and contain clusters of dramatically different sizes and morphologies. We also determine the diffusion spreadability ${\cal S}(t)$ and estimate the fluid permeability $k$ associated with all of the constructed materials directly from the corresponding ${\tilde χ}_{_V}({k})$ functions. We find that varying the length-scale parameter within each class of ${\tilde χ}_{_V}({k})$ functions can also lead to orders of magnitude variation of ${\cal S}(t)$ at intermediate and long time scales. Moreover, we find that increasing solid volume fraction $ϕ_1$ and correlation length $a$ in the constructed media generally leads to a decrease in the dimensionless fluid permeability $k/a^2$. These results indicate the feasibility of employing parameterized ${\tilde χ}_{_V}({k})$ for designing composites with targeted transport properties.

cond-mat.mtrl-sci↗

arXiv2024

Unlocking TriLevel Learning with Level-Wise Zeroth Order Constraints: Distributed Algorithms and Provable Non-Asymptotic Convergence

Trilevel learning (TLL) found diverse applications in numerous machine learning applications, ranging from robust hyperparameter optimization to domain adaptation. However, existing researches primarily focus on scenarios where TLL can be addressed with first order information available at each level, which is inadequate in many situations involving zeroth order constraints, such as when black-box models are employed. Moreover, in trilevel learning, data may be distributed across various nodes, necessitating strategies to address TLL problems without centralizing data on servers to uphold data privacy. To this end, an effective distributed trilevel zeroth order learning framework DTZO is proposed in this work to address the TLL problems with level-wise zeroth order constraints in a distributed manner. The proposed DTZO is versatile and can be adapted to a wide range of (grey-box) TLL problems with partial zeroth order constraints. In DTZO, the cascaded polynomial approximation can be constructed without relying on gradients or sub-gradients, leveraging a novel cut, i.e., zeroth order cut. Furthermore, we theoretically carry out the non-asymptotic convergence rate analysis for the proposed DTZO in achieving the $ε$-stationary point. Extensive experiments have been conducted to demonstrate and validate the superior performance of the proposed DTZO, e.g., it approximately achieves up to a 40$\%$ improvement in performance.

cs.LG↗

arXiv2024

Structural Properties of Hyperuniform Networks

Disordered hyperuniform many-particle systems are recently discovered exotic states of matter, characterized by a complete suppression of normalized infinite-wavelength density fluctuations and lack of conventional long-range order. Here, we begin a program to quantify the structural properties of nonhyperuniform and hyperuniform networks. In particular, large two-dimensional (2D) Voronoi networks (graphs) containing approximately 10,000 nodes are created from a variety of different point configurations, including the antihyperuniform HIP, nonhyperuniform Poisson process, nonhyperuniform RSA saturated packing, and both non-stealthy and stealthy hyperuniform point processes. We carry out an extensive study of the Voronoi-cell area distribution of each of the networks through determining multiple metrics that characterize the distribution, including their higher-cumulants. We show that the HIP distribution is far from Gaussian; the Poisson and non-stealthy hyperuniform distributions are Gaussian-like distributions, the RSA and the highest stealthy hyperuniform distributions are also non-Gaussian, with diametrically opposite non-Gaussian behavior of the HIP. Moreover, we compute the Voronoi-area correlation functions $C_{00}(r)$ for the networks [M. A. Klatt and S. Torquato, Phys. Rev. E {\bf 90}, 052120 (2014)]. We show that the correlation functions $C_{00}(r)$ qualitatively distinguish the antihyperuniform, nonhyperuniform and hyperuniform Voronoi networks. We find strong anticorrelations in $C_{00}(r)$ (i.e., negative values) for the hyperuniform networks.

cond-mat.stat-mech↗

arXiv2024

Tri-Level Navigator: LLM-Empowered Tri-Level Learning for Time Series OOD Generalization

Out-of-Distribution (OOD) generalization in machine learning is a burgeoning area of study. Its primary goal is to enhance the adaptability and resilience of machine learning models when faced with new, unseen, and potentially adversarial data that significantly diverges from their original training datasets. In this paper, we investigate time series OOD generalization via pre-trained Large Language Models (LLMs). We first propose a novel \textbf{T}ri-level learning framework for \textbf{T}ime \textbf{S}eries \textbf{O}OD generalization, termed TTSO, which considers both sample-level and group-level uncertainties. This formula offers a fresh theoretic perspective for formulating and analyzing OOD generalization problem. In addition, we provide a theoretical analysis to justify this method is well motivated. We then develop a stratified localization algorithm tailored for this tri-level optimization problem, theoretically demonstrating the guaranteed convergence of the proposed algorithm. Our analysis also reveals that the iteration complexity to obtain an $ε$-stationary point is bounded by O($\frac{1}{ε^{2}}$). Extensive experiments on real-world datasets have been conducted to elucidate the effectiveness of the proposed method.

cs.LG↗

arXiv2024

Optimization of a lattice spring model with elastoplastic conducting springs: A case study

We consider a simple lattice spring model in which every spring is elastoplastic and is capable to conduct current. The elasticity bounds of spring $i$ are taken as $[-c_i,c_i]$ and the resistance of spring $i$ is taken as $1/c_i$, which allows us to compute the resistance of the system. The model is further subjected to a gradual stretching and, due to plasticity, the response force increases until a certain terminal value. We demonstrate that the recently developed sweeping process theory can be used to optimize the interplay between the terminal response force and the resistance on a physical domain of parameters $c_i.$ The proposed methodology can be used by practitioners for the design of multi-functional materials as an alternative to topological optimization.

math.OC↗

arXiv2024

EAGLE: Towards Efficient Arbitrary Referring Visual Prompts Comprehension for Multimodal Large Language Models

Recently, Multimodal Large Language Models (MLLMs) have sparked great research interests owing to their exceptional content-reasoning and instruction-following capabilities. To effectively instruct an MLLM, in addition to conventional language expressions, the practice of referring to objects by painting with brushes on images has emerged as a prevalent tool (referred to as "referring visual prompts") due to its efficacy in aligning the user's intention with specific image regions. To accommodate the most common referring visual prompts, namely points, boxes, and masks, existing approaches initially utilize specialized feature encoding modules to capture the semantics of the highlighted areas indicated by these prompts. Subsequently, these encoded region features are adapted to MLLMs through fine-tuning on a meticulously curated multimodal instruction dataset. However, such designs suffer from redundancy in architecture. Moreover, they face challenges in effectively generalizing when encountering a diverse range of arbitrary referring visual prompts in real-life scenarios. To address the above issues, we propose EAGLE, a novel MLLM that empowers comprehension of arbitrary referring visual prompts with less training efforts than existing approaches. Specifically, our EAGLE maintains the innate format of the referring visual prompts as colored patches rendered on the given image for conducting the instruction tuning. Our approach embeds referring visual prompts as spatial concepts conveying specific spatial areas comprehensible to the MLLM, with the semantic comprehension of these regions originating from the MLLM itself. Besides, we also propose a Geometry-Agnostic Learning paradigm (GAL) to further disentangle the MLLM's region-level comprehension with the specific formats of referring visual prompts. Extensive experiments are conducted to prove the effectiveness of our proposed method.

cs.CV↗

arXiv2024

Automated Quantification of White Blood Cells in Light Microscopic Images of Injured Skeletal Muscle

White blood cells (WBCs) are the most diverse cell types observed in the healing process of injured skeletal muscles. In the course of healing, WBCs exhibit dynamic cellular response and undergo multiple protein expression changes. The progress of healing can be analyzed by quantifying the number of WBCs or the amount of specific proteins in light microscopic images obtained at different time points after injury. In this paper, we propose an automated quantifying and analysis framework to analyze WBCs using light microscopic images of uninjured and injured muscles. The proposed framework is based on the Localized Iterative Otsu's threshold method with muscle edge detection and region of interest extraction. Compared with the threshold methods used in ImageJ, the LI Otsu's threshold method has high resistance to background area and achieves better accuracy. The CD68-positive cell results are presented for demonstrating the effectiveness of the proposed work.

eess.IV↗

arXiv2024

Universal Hyperuniform Organization in Looped Leaf Vein Networks

Leaf vein network is a hierarchical vascular system that transports water and nutrients to the leaf cells. The thick primary veins form a branched network, while the secondary veins develop closed circuits forming a well-defined cellular structure. Through extensive analysis of a variety of distinct leaf species, we discover that the apparently disordered cellular structures of the secondary vein networks exhibit a universal hyperuniform organization and possess a hidden order on large scales. Disorder hyperuniform (DHU) systems lack conventional long-range order, yet they completely suppress normalized large-scale density fluctuations like crystals. Specifically, we find that the distributions of the geometric centers associated with the vein network loops possess a vanishing static structure factor in the zero-wavenumber limit, i.e., $S(k) \sim k^α$, where $α\approx 0.64$, providing an example of class III hyperuniformity in biology. This hyperuniform organization leads to superior efficiency of diffusive transport, as evidenced by the much faster convergence of the time-dependent spreadability $\mathcal{S}(t)$ to its long-time asymptotic limit, compared to that of other uncorrelated or correlated disordered but non-hyperuniform organizations. Our results also have implications for the discovery and design of novel disordered network materials with optimal transport properties.

cond-mat.soft↗