Searcharxiv⌕ Search

arXiv subjects

Yangyang Xu

Publications and source records attributed to Yangyang Xu.

At least 19 recordsLinked to original sources
arXiv2026

Task-State Adaptation with Prototype Memory for Multi-Task Dense Prediction

Vision foundation backbones provide strong representations for dense prediction, yet a single shared feature still needs to support tasks with different, image-dependent adaptation requirements. We propose MemMTL, a multi-task dense prediction framework that estimates a compact task state from global visual context and refines it through a learnable task-state prototype memory. The refined state is converted into task-conditioned expert logits and combined with token-level logits before sparse top-$k$ selection over a local expert bank shared by all tasks. A separate task-agnostic residual bank provides a common adaptation path, and both paths are added once to the backbone feature before task-specific prediction. We specify a matched evaluation protocol on NYUD-v2 and PASCAL-Context with SAM 3 and ViT-L backbones to measure predictive quality, computational cost, and the contributions of task-state conditioning, prototype retrieval, and sparse routing. The numerical record in the present working draft predates this canonical implementation and must be regenerated before it can support empirical claims.

cs.CV↗
arXiv2026

Scalable Dynamic Optimal Transport via Distributed Linearized ADMM

In this paper, we address two fundamental challenges in the numerical solution of dynamic opti- mal transport (OT) problems. The first challenge arises when the initial and/or terminal densities approach zero and no positive lower bound is available. In this regime, conventional methods may become unstable or computationally inefficient, since the Lipschitz constant of the objective can scale like the reciprocal of the cube of the density. As a result, near-zero regions may lead to slow convergence or even divergence. The second challenge concerns the substantial memory cost of the dynamic formulation, whose discretization over fine spatial and temporal grids requires storing vari- ables across the entire space-time domain. This storage burden quickly becomes prohibitive as the grid is refined or the spatial dimension increases. To overcome the first difficulty, we reformulate the classical discretized dynamic OT problem so that the resulting objective admits an exact proximal mapping. When combined with a linearized alternating direction method of multipliers (LADMM), this reformulation yields an efficient and robust algorithm that remains stable even in challenging settings where the density may vanish. To reduce the memory burden, we further introduce a dis- tributed formulation in which the optimization variables are partitioned across multiple agents. This design substantially lowers the storage requirement for each agent and can also lead to computational acceleration. We validate the proposed framework through numerical experiments in one- and two- dimensional spatial settings under varying levels of difficulty. The results demonstrate the stability, robustness, and scalability of the proposed method in comparison with conventional approaches.

math.OC↗
arXiv2026

GateMem: Benchmarking Memory Governance in Multi-Principal Shared-Memory Agents

Memory benchmarks for LLM agents largely assume single-user settings, leaving shared assistants for hospitals, workplaces, campuses, and households understudied. In these deployments, multiple principals write to a common memory pool and query it under different roles, scopes, and relationships, so memory quality requires governance as well as recall. We introduce GateMem, a benchmark for multi-principal shared-memory agents. GateMem jointly evaluates utility for legitimate long-horizon requests with state updates, access control across contextual authorization boundaries, and agent-facing active forgetting after explicit deletion requests. It spans medical, office, education, and household domains, with long-form multi-party episodes, incremental memory injection, hidden checkpoints, structured judging, and leak-target annotations. Across diverse baselines and backbone models, no method simultaneously achieves strong utility, robust access control, and reliable forgetting. Long-context prompting often yields the best governance score at high token cost, while retrieval-based and external-memory methods reduce cost yet still leak unauthorized or deleted information. These results show current memory agents remain far from reliable shared institutional deployment.

cs.LG↗
arXiv2026

First-Order Methods for Solving Convex (Strongly) Concave Minimax Problems with Functional Constraints

Minimax problems arise in many applications, including robust learning and Stackelberg models. Most existing methods for minimax problems address unconstrained or projection-friendly settings, while functional constrained minimax problems remain far less explored. We study a class of convex-(strongly-)concave minimax problems with functional constraints. By exploiting strong duality, we incorporate the inner-maximization functional constraints into the objective. This allows us to efficiently obtain inexact gradients of the primal function of the reformulation and to design a proximal augmented Lagrangian method (PALM). Each PALM subproblem is solved by an inexact accelerated proximal gradient scheme to handle inexact gradients arising from approximately solving an auxiliary maximization subproblem. We show that the proposed method returns an $\varepsilon$-KKT point and a primal $\varepsilon$-optimal solution, with $\tilde{\mathcal{O}}(\varepsilon^{-1})$ first-order oracle and iteration complexity in the convex-strongly-concave case. For the convex-concave case, the complexity remains the same for the primal gradient evaluation but increases to $\tilde{\mathcal{O}}(\varepsilon^{-\frac{3}{2}})$ for the dual part.

math.OC↗
arXiv2026

MemSlides: A Hierarchical Memory Driven Agent Framework for Personalized Slide Generation with Multi-turn Local Revision

Personalized presentation generation requires more than conditioning on a current prompt or template: agents must preserve stable user preferences across tasks, retain newly introduced preferences and constraints during multi-turn revision, and carry out local edits reliably. We propose MemSlides, a hierarchical memory framework for personalized presentation agents that separates long-term memory from working memory and further divides long-term memory into user profile memory and tool memory. User profile memory stores intent-conditioned profiles for round-0 personalization, working memory carries active preferences and session constraints across revision rounds, and tool memory stores reusable execution experience for reliable localized editing. MemSlides pairs this memory design with scoped slide-local revision, so targeted updates act on the smallest affected region instead of repeatedly regenerating the full deck. In controlled experiments, user profile memory improves persona-alignment judgments on a multi-persona, multi-intent profile bank, tool-memory injection improves closed-loop modify behavior in diagnostic matched-pair settings, and qualitative cases illustrate working memory's ability to carryover preferences. Taken together, these results suggest that effective personalization in presentation authoring depends on separating persistent user profiles, session-level working memory, and reusable execution experience across generation and localized revision.

cs.CL↗
arXiv2026

$h$-control: Training-Free Camera Control via Block-Conditional Gibbs Refinement

Training-free camera control for pretrained flow-matching video generators is a partial-observation inverse problem: a depth-warped guidance video supplies noisy evidence on a subset of latent sites, which the sampler must reconcile with the pretrained prior. Existing methods struggle to balance the trade-off between trajectory adherence and visual quality and the heuristic guidance-strength tuning lacks robustness. We propose \textbf{$h$-control}, which resolves this dilemma through a structural change to the sampler: each outer hard-replacement guidance step is augmented with an inner-loop \emph{block-conditional pseudo-Gibbs refinement} on the unobserved complement at the same noise level, with provable convergence to the partial-observation conditional data law. To accelerate convergence on high-dimensional video latents, we exploit their conditional locality, partitioning the unobserved complement into 3D patches, each tracked by a custom mixing indicator that adaptively freezes converged patches. On RealEstate10K and DAVIS, \textbf{$h$-control} attains the best FVD against all seven training-free and training-based competitors, outperforming every training-free baseline on every reported metric.

cs.CV↗
arXiv2026

SAT: Sequential Agent Tuning for Coordinator Free Plug and Play Multi-LLM Training with Monotonic Improvement Guarantees

Large language models (LLMs) with a large number of parameters achieve strong performance but are often prohibitively expensive to deploy. Recent work explores using teams of smaller, more efficient LLMs that collectively match or even outperform a single large model. However, jointly updating multiple agents introduces compounding distribution shifts, making coordination and stability during training difficult. We address this by introducing Sequential Agent Tuning (SAT), a coordinator-free training paradigm. SAT represents the team as a factorized policy and employs block-coordinate updates over agents, enabling scalable, decentralized training without a central controller. Specifically, we develop a sequence-aware, on-policy advantage estimator that conditions on the evolving team policy, coupled with per-agent KL trust regions that isolate occupancy drift. Theoretically, this framework provides two critical guarantees. First, it ensures monotonic improvement, stabilizing the training process. Second, it establishes provable plug-and-play invariance: any agent can be upgraded to a stronger model without retraining the rest of the team, with a formal guarantee that the performance bound improves. Empirically, a team of three 4B agents (12B total) trained with SAT surpasses the much larger Qwen3-32B on AIME24/25 benchmarks by 3.9\% on average. We validate our plug-and-play theory by swapping in two 8B agents, which boosts the composite score by 10.4\%. We provide code and appendix of proof at https://github.com/Yydc/SAT-AAMAS

cs.LG↗
arXiv2026

LoDAdaC: a unified local training-based decentralized framework with adaptive gradients and compressed communication

In the decentralized distributed learning, achieving fast convergence and low communication cost is essential for scalability and high efficiency. Adaptive gradient methods, such as Adam, have demonstrated strong practical performance in deep learning and centralized distributed settings. However, their convergence properties remain largely unexplored in decentralized settings involving multiple local training steps, such as federated learning. To address this limitation, we propose LoDAdaC, a unified multiple Local Training (MLT) Decentralized framework with Adam-type updates and Compressed communication (CC). LoDAdaC accommodates a broad class of optimizers for its local adaptive updates, including AMSGrad, Adam, and AdaGrad; it is compatible with standard (possibly biased) compressors such as low-bit quantization and sparsification. MLT and CC enable LoDAdaC to achieve multiplied reduction of communication cost, while the technique of adaptive updates enables fast convergence. We rigorously prove the combined advantage through complexity analysis. In addition, experiments on image classification and GPT-style language model training validate our theoretical findings and show that LoDAdaC significantly outperforms existing decentralized algorithms in terms of convergence speed and communication efficiency.

cs.LG↗
arXiv2026

NimbusGS: Unified 3D Scene Reconstruction under Hybrid Weather

We present NimbusGS, a unified framework for reconstructing high-quality 3D scenes from degraded multi-view inputs captured under diverse and mixed adverse weather conditions. Unlike existing methods that target specific weather types, NimbusGS addresses the broader challenge of generalization by modeling the dual nature of weather: a continuous, view-consistent medium that attenuates light, and dynamic, view-dependent particles that cause scattering and occlusion. To capture this structure, we decompose degradations into a global transmission field and per-view particulate residuals. The transmission field represents static atmospheric effects shared across views, while the residuals model transient disturbances unique to each input. To enable stable geometry learning under severe visibility degradation, we introduce a geometry-guided gradient scaling mechanism that mitigates gradient imbalance during the self-supervised optimization of 3D Gaussian representations. This physically grounded formulation allows NimbusGS to disentangle complex degradations while preserving scene structure, yielding superior geometry reconstruction and outperforming task-specific methods across diverse and challenging weather conditions. Code is available at https://github.com/lyy-ovo/NimbusGS.

cs.CV↗
arXiv2026

SHIFT: Motion Alignment in Video Diffusion Models with Adversarial Hybrid Fine-Tuning

Image-conditioned video diffusion models achieve impressive visual realism but often suffer from weakened motion fidelity, e.g., reduced motion dynamics or degraded long-term temporal coherence, especially after fine-tuning. We study motion alignment in video diffusion models post-training. To address this, we introduce pixel-motion rewards based on pixel flux dynamics, capturing both instantaneous and long-term motion consistency. We further propose \underline{S}mooth \underline{H}ybr\underline{i}d \underline{F}ine-\underline{t}uning (SHIFT), a scalable reward-driven framework that unifies supervised fine-tuning and advantage-weighted fine-tuning. Benefiting from novel adversarial advantages, SHIFT improves convergence speed and mitigates reward hacking. Experiments show that our approach efficiently resolves dynamic-degree collapse in modern video diffusion models supervised fine-tuning. Project page: https://xiye20.github.io/projects/SHIFT/.

cs.CV↗
arXiv2026

Content-Aware Frequency Encoding for Implicit Neural Representations with Fourier-Chebyshev Features

Implicit Neural Representations (INRs) have emerged as a powerful paradigm for various signal processing tasks, but their inherent spectral bias limits the ability to capture high-frequency details. Existing methods partially mitigate this issue by using Fourier-based features, which usually rely on fixed frequency bases. This forces multi-layer perceptrons (MLPs) to inefficiently compose the required frequencies, thereby constraining their representational capacity. To address this limitation, we propose Content-Aware Frequency Encoding (CAFE), which builds upon Fourier features through multiple parallel linear layers combined via a Hadamard product. CAFE can explicitly and efficiently synthesize a broader range of frequency bases, while the learned weights enable the selection of task-relevant frequencies. Furthermore, we extend this framework to CAFE+, which incorporates Chebyshev features as a complementary component to Fourier bases. This combination provides a stronger and more stable frequency representation. Extensive experiments across multiple benchmarks validate the effectiveness and efficiency of our approach, consistently achieving superior performance over existing methods. Our code is available at https://github.com/JunboKe0619/CAFE.

cs.CV↗
arXiv2026

Reparameterized Tensor Ring Functional Decomposition for Multi-Dimensional Data Recovery

Tensor Ring (TR) decomposition is a powerful tool for high-order data modeling, but is inherently restricted to discrete forms defined on fixed meshgrids. In this work, we propose a TR functional decomposition for both meshgrid and non-meshgrid data, where factors are parameterized by Implicit Neural Representations (INRs). However, optimizing this continuous framework to capture fine-scale details is intrinsically difficult. Through a frequency-domain analysis, we demonstrate that the spectral structure of TR factors determines the frequency composition of the reconstructed tensor and limits the high-frequency modeling capacity. To mitigate this, we propose a reparameterized TR functional decomposition, in which each TR factor is a structured combination of a learnable latent tensor and a fixed basis. This reparameterization is theoretically shown to improve the training dynamics of TR factor learning. We further derive a principled initialization scheme for the fixed basis and prove the Lipschitz continuity of our proposed model. Extensive experiments on image inpainting, denoising, super-resolution, and point cloud recovery demonstrate that our method achieves consistently superior performance over existing approaches. Code is available at https://github.com/YangyangXu2002/RepTRFD.

cs.CV↗
arXiv2026

A variance reduced framework for (non)smooth nonconvex-nonconcave stochastic minimax problems with extended Kurdyka-Lojasiewicz property

In this paper, we study stochastic constrained minimax optimization problems with nonconvex-nonconcave structure, a central problem in modern machine learning, for which reliable and efficient algorithms remain largely unexplored due to its inherent challenges. Prior approaches for nonconvex minimax optimization often require (strong) concavity on the maximization part, or certain restrictive geometric assumptions on the joint objective to have guaranteed convergence. In contrast, our method only assumes weak convexity in the primal variable and the extended Kurdyka-Lojasiewicz (KL) property, with exponent $\theta \in [0,1]$, in the dual variable, significantly broadening the class of tractable problems. To this end, we propose a variance reduced algorithm that provably handles this general setting and achieves an $\varepsilon$-stationary solution with state-of-the-art sample complexity: in the smooth finite-sum setting, the sample complexity is $\mathcal{O}\left(\sqrt{N}\,\varepsilon^{-\max\{4\theta,2\}}\right)$, where $N$ is the number of total samples, and in the online smooth setting, it is $\mathcal{O}\Big(\varepsilon^{-\max\{6\theta,3\}}\Big)$. For the structured nonsmooth problem, the sample complexity is $\mathcal{O}\left(\sqrt{N}\,\max\Big\{\varepsilon^{-3}, \varepsilon^{-5\theta}, \varepsilon^{-\frac{11\theta-3}{2\theta}}\Big\}\right)$ and $\mathcal{O}\left(\max\left\{\varepsilon^{-4}, \varepsilon^{-\frac{15\theta-1}{2}}, \varepsilon^{-\frac{31\theta-9}{4\theta}}\right\}\right)$ respectively for the two settings. To the best of our knowledge, this is the first unified framework that jointly accommodates weak convexity, the extended KL property, and variance-reduced stochastic updates, making it highly suitable for large-scale applications.

math.OC↗
arXiv2026

OmniVTON++: Training-Free Universal Virtual Try-On with Principal Pose Guidance

Image-based Virtual Try-On (VTON) concerns the synthesis of realistic person imagery through garment re-rendering under human pose and body constraints. In practice, however, existing approaches are typically optimized for specific data conditions, making their deployment reliant on retraining and limiting their generalization as a unified solution. We present OmniVTON++, a training-free VTON framework designed for universal applicability. It addresses the intertwined challenges of garment alignment, human structural coherence, and boundary continuity by coordinating Structured Garment Morphing for correspondence-driven garment adaptation, Principal Pose Guidance for step-wise structural regulation during diffusion sampling, and Continuous Boundary Stitching for boundary-aware refinement, forming a cohesive pipeline without task-specific retraining. Experimental results demonstrate that OmniVTON++ achieves state-of-the-art performance across diverse generalization settings, including cross-dataset and cross-garment-type evaluations, while reliably operating across scenarios and diffusion backbones within a single formulation. In addition to single-garment, single-human cases, the framework supports multi-garment, multi-human, and anime character virtual try-on, expanding the scope of virtual try-on applications. The code is available at https://github.com/Jerome-Young/OmniVTON-PlusPlus.

cs.CV↗
arXiv2026

Alternating Direction Method of Multipliers for nonlinear constrained convex problems and applications to distributed resource allocation and constrained machine learning

We study a class of structured convex optimization problems, which have a two-block separable objective and nonlinear functional constraints as well as affine constraints that couple the two block variables. Such problems naturally arise from distributed resource allocation and constrained machine learning. To achieve high communication efficiency for the distributed applications, we propose a nonlinear alternating direction method of multipliers (NL-ADMM) that preserves the classical splitting structure while accommodating general convex functional constraints. Unlike existing ADMM variants for nonconvex constrained problems, the proposed method does not require smoothness of the objective functions or differentiability of the constraint mapping, by leveraging convexity of the considered problem. We establish global convergence and an ergodic $O(1/k)$ convergence rate of NL-ADMM by assuming the existence of a KKT solution. The results extend those of ADMM for linearly constrained convex problems. Numerical experiments are conducted on two representative distributed tasks. The results on numerous instances demonstrate that NL-ADMM can achieve (in many cases) 100x higher communication efficiency than the classic augmented Lagrangian method and nearly 2x higher than the Douglas-Rachford operator splitting method, making the new method well suited for large-scale distributed learning systems.

math.OC↗
arXiv2026

MotionAdapter: Video Motion Transfer via Content-Aware Attention Customization

Recent advances in diffusion-based text-to-video models, particularly those built on the diffusion transformer architecture, have achieved remarkable progress in generating high-quality and temporally coherent videos. However, transferring complex motions between videos remains challenging. In this work, we present MotionAdapter, a content-aware motion transfer framework that enables robust and semantically aligned motion transfer within DiT-based video diffusion models. Our key insight is that effective motion transfer requires 1) explicit disentanglement of motion from appearance and 2) adaptive customization of motion to target content. MotionAdapter first isolates motion by analyzing cross-frame attention within 3D full-attention modules to extract attention-derived motion fields. To bridge the semantic gap between reference and target videos, we further introduce a DINO-guided motion customization module that rearranges and refines motion fields based on content correspondences. The customized motion field is then used to guide the DiT denoising process, ensuring that the synthesized video inherits the reference motion while preserving target appearance and semantics. Extensive experiments demonstrate that MotionAdapter outperforms state-of-the-art methods in both qualitative and quantitative evaluations. Moreover, MotionAdapter naturely support complex motion transfer and motion editing tasks such as zooming in/out and composition.

cs.CV↗
arXiv2025

DeshadowMamba: Deshadowing as 1D Sequential Similarity

Recent deep models for image shadow removal often rely on attention-based architectures to capture long-range dependencies. However, their fixed attention patterns tend to mix illumination cues from irrelevant regions, leading to distorted structures and inconsistent colors. In this work, we revisit shadow removal from a sequence modeling perspective and explore the use of Mamba, a selective state space model that propagates global context through directional state transitions. These transitions yield an efficient global receptive field while preserving positional continuity. Despite its potential, directly applying Mamba to image data is suboptimal, since it lacks awareness of shadow-non-shadow semantics and remains susceptible to color interference from nearby regions. To address these limitations, we propose CrossGate, a directional modulation mechanism that injects shadow-aware similarity into Mamba's input gate, allowing selective integration of relevant context along transition axes. To further ensure appearance fidelity, we introduce ColorShift regularization, a contrastive learning objective driven by global color statistics. By synthesizing structured informative negatives, it guides the model to suppress color contamination and achieve robust color restoration. Together, these components adapt sequence modeling to the structural integrity and chromatic consistency required for shadow removal. Extensive experiments on public benchmarks demonstrate that DeshadowMamba achieves state-of-the-art visual quality and strong quantitative performance.

cs.CV↗
arXiv2025

A Depth-Independent Linear Chain Ansatz for Large-Scale Quantum Approximate Optimization

Combinatorial optimization lies at the heart of numerous real-world applications. For a broad category of optimization problems, quantum computing is expected to exhibit quantum speed-up over classic computing. Among various quantum algorithms, the Quantum Approximate Optimization Algorithm (QAOA), as one of variational quantum algorithms, shows promise on demonstrating quantum advantage on noisy intermediate-scale quantum (NISQ) hardware. However, with increasing problem size, the circuit depth demanded by original QAOA scales rapidly and quickly surpasses the threshold at which meaningful results can be obtained. To address this challenge, in this work, we propose a variant of QAOA (termed linear chain QAOA) and demonstrate its advantages over original QAOA on paradigmatic MaxCut problems. In original QAOA, each graph edge is encoded with one entangling gate. In our ansatz, we locate a linear chain from the original MaxCut graph and place entangling gates sequentially along this chain. This linear-chain ansatz is featured by shallow quantum circuits and with the low execution time that scales independently of the problem size. Leveraging this ansatz, we demonstrate an approximation ratio of 0.78 (without post-processing) on non-hardware-native random regular MaxCut instances with 100 vertices in a digital quantum processor using 100 qubits. Our findings offer new insights into the design of hardware-efficient ansatz and point toward a promising route for tackling large-scale combinatorial optimization problems on NISQ devices.

quant-ph↗