arXiv subjects

Xu Zhang

Publications and source records attributed to Xu Zhang.

At least 91 records · Page 5Linked to original sources

arXiv2026

Thresholded Cross-Attention for Reliable Intensity-Chromaticity Fusion in Low-Light Image Enhancement

Low-Light Image Enhancement (LLIE) requires a careful balance among noise suppression, color fidelity, and efficiency. Recent HVI-based methods alleviate color entanglement by decoupling intensity and chromaticity, yet how reliably the two streams are fused again is an overlooked factor that largely determines the final quality. We observe that the confidence of cross-stream attention is strongly layer-dependent, so the fixed-quota selection of Top-K sparse attention is mismatched to it, discarding informative dependencies in some layers while retaining noisy ones in others. Motivated by this observation, we propose TCA-Net, a network built around Thresholded Cross-Attention that targets reliable intensity-chromaticity fusion in the HVI space rather than introducing yet another color representation. At its core, TCA replaces the rigid Top-K quota with a fixed confidence threshold whose retained cardinality is input- and layer-adaptive, retaining only high-confidence cross-stream interactions while suppressing unreliable ones. Around this core, two complementary designs clean up the fusion before and after it: a Phase-guided Fourier Interaction Module provides a structure-aware brightness initialization for the intensity stream prior to fusion, and a Decoupled Dual-Stream Guidance Module constructs residual intensity features to suppress chromaticity leakage during reconstruction. A Scale-Aware Consistency Regularization further improves structural robustness under scale perturbations during training. Extensive experiments on LOL-v1, LOL-v2, Sony-Total-Dark, and LSRW-Huawei demonstrate that TCA-Net delivers competitive restoration accuracy, improved color fidelity, and a compact parameter size.

cs.CV↗

arXiv2026

An efficient method based on the evolutionary center algorithm for optimizing chemical-diffusive models for flame acceleration and DDT

This paper presents an efficient method based on Evolutionary Center Algorithm (ECA) for accurately and efficiently determining the optimal reaction and diffusion parameters for Chemical-Diffusive Models (CDM) to simulate flame acceleration (FA) and deflagration-to-detonation transition (DDT). The proposed method leverages the global search capability of the ECA and the local optimization strength of the Nelder-Mead (NM) algorithm. The hybrid approach (ECA-NM) can efficiently optimize CDM parameters that are capable of accurately reproducing the major properties of combustion waves. The CDMs for premixed flames and detonations of hydrogen in air or oxygen were developed using the present ECA-NM method and validated against canonical tests of combustion waves and previous experiments of FA and DDT. The results show that the major flame and detonation properties calculated using the developed CDMs match those obtained from detailed chemical reaction mechanisms over a wide range of equivalence ratio. The simulated FA and DDT in a channel also agree qualitatively and quantitatively with experiments in terms of complex flame instabilities (e.g., tulip and distorted tulip flames), flame displacement speed, and detonation occurrence. In addition, detailed comparisons to the traditional genetic algorithm demonstrate that the developed ECA-NM method diminishes the global error by four orders of magnitude while reducing the computational cost by two orders of magnitude. This work provides a significantly efficient method for developing chemical-diffusive models that allows quantitative multi-scale simulations of transient flames and detonations in complex scenarios.

physics.chem-ph↗

arXiv2026

PIT-SUN: A Deployable Empirical Marginal Transform Framework with Expectation-Consistent Recovery for Regression in Recommender Systems

Estimating original-space conditional expectations is central to value-driven recommender systems, including dwell time, GMV, and LTV forecasting. Standard MSE is expectation-consistent in principle, but its gradients become unstable on heavy-tailed, zero-inflated, and multimodal targets, causing mean collapse and tail shrinkage. Target transformation alleviates this scale conflict, yet any useful nonlinear marginal transform loses expectation consistency under direct inversion. This is not an implementation oversight: a direct inverse-transform estimator is universally expectation-consistent only when the inverse transform is affine, which cannot simultaneously provide bounded tail compression. Existing conditionally linear recovery methods restore expectation consistency, but still leave open which coordinate, inverse lookup, recovery base, and deployment monitor should be selected for sparse complex marginals. We propose \textbf{P}robability-\textbf{I}ntegral-\textbf{TranS}formed \textbf{Un}biased recovery (\textbf{PIT-SUN}), a deployable empirical marginal recovery framework. PIT-SUN uses one empirical marginal table to define a bounded normal-score coordinate, its inverse-quantile lookup, a variance-controlled recovery base, and drift monitoring, then applies multiplicative SUN recovery to estimate the original-space expectation instead of directly inverting transformed predictions. Experiments on synthetic distributions, public benchmarks, large-scale industrial datasets, and online deployment show robust improvements in point accuracy, calibration, and ranking quality with lightweight deployment overhead.

cs.LG↗

arXiv2026

What Memory Do GUI Agents Really Need? From Passive Records to Active Task-Driving States

Mobile GUI agents increasingly face long-horizon tasks that require reading, updating, and reusing task-relevant data across pages and applications. Existing methods treat memory largely as passive storage, where past observations are accumulated and retrieved when needed. Yet retrieving a value does not reveal its current role in the workflow. The agent must still infer from accumulated records whether the value should be used now, has already been used, or must wait for a later dependency. This implicit reconstruction becomes unreliable in long trajectories with repeated values, distractors, and outdated states, causing repeated or missed operations. To address this, we propose Active Task Driving Memory (ATMem), which shifts GUI-agent memory from passive storage to an actively maintained execution state. ATMem maintains task-relevant information as a continually updated execution state that links each value to its role and current status, enabling action selection based on the current workflow state. While supervised fine-tuning enables the agent to construct ATMem, it does not teach when ATMem is beneficial. We therefore introduce STR-GRPO, an online reinforcement learning method that encourages selective use of ATMem based on its contribution to task completion. STR-GRPO contrasts memory-on and memory-off rollouts to estimate when memory use improves execution, while memory-cost-aware reward discourages costly memory usage that does not improve execution. To evaluate whether agents can complete all in-scope work while avoiding out-of-scope actions, we build a challenging mobile benchmark. From a list of near identical entries, agents must act on every entry that satisfies the instruction and reject entries that violate its constraints. We further introduce App-Level Progress and Scope-Aware F1 to measure these two dimensions separately.

cs.CV↗

arXiv2026

Diff-MN: Diffusion Parameterized MoE-NCDE for Continuous Time Series Generation with Irregular Observations

Time series generation (TSG) is widely used across domains, yet most existing methods assume regular sampling and fixed output resolutions. These assumptions are often violated in practice, where observations are irregular and sparse, while downstream applications require continuous and high-resolution TS. Although Neural Controlled Differential Equation (NCDE) is promising for modeling irregular TS, it is constrained by a single dynamics function, tightly coupled optimization, and limited ability to adapt learned dynamics to newly generated samples from the generative model. We propose Diff-MN, a continuous TSG framework that enhances NCDE with a Mixture-of-Experts (MoE) dynamics function and a decoupled architectural design for dynamics-focused training. To further enable NCDE to generalize to newly generated samples, Diff-MN employs a diffusion model to parameterize the NCDE temporal dynamics parameters (MoE weights), i.e., jointly learn the distribution of TS data and MoE weights. This design allows sample-specific NCDE parameters to be generated for continuous TS generation. Experiments on ten public and synthetic datasets demonstrate that Diff-MN consistently outperforms strong baselines on both irregular-to-regular and irregular-to-continuous TSG tasks. The code is available at the link https://github.com/microsoft/TimeCraft/tree/main/Diff-MN.

cs.LG↗

arXiv2026

Uniqueness of bound states for sublinear elliptic equations

We investigate the uniqueness of radial bound state solutions to the sublinear elliptic equation \[ \begin{cases} -Δu - u + |u|^{q-2}u = 0 & \text{in } \mathbb{R}^n,\cr u(x) \to 0 & \text{as } |x| \to \infty, \end{cases} \] where $q\in(1,2)$ and $n\geq 2$. A distinctive feature of this problem is the non-Lipschitz singularity of the nonlinearity at the origin, which gives rise to compactly supported ground states and bound states. Using a shooting argument together with a detailed analysis of the linearized variation with respect to the initial value, we prove that for every prescribed integer $k\geq 1$, the equation admits exactly one radial bound state solution with $k$ simple zeros, up to sign reflection and spatial translation. In addition, our analysis yields a classification of radial solutions according to the initial value and describes their behavior near the finite support boundary.

math.AP↗

arXiv2026

SEMixer: Semantics Enhanced MLP-Mixer for Multiscale Mixing and Long-term Time Series Forecasting

Modeling multiscale patterns is crucial for long-term time series forecasting (TSF). However, redundancy and noise in time series, together with semantic gaps between non-adjacent scales, make the efficient alignment and integration of multi-scale temporal dependencies challenging. To address this, we propose SEMixer, a lightweight multiscale model designed for long-term TSF. SEMixer features two key components: a Random Attention Mechanism (RAM) and a Multiscale Progressive Mixing Chain (MPMC). RAM captures diverse time-patch interactions during training and aggregates them via dropout ensemble at inference, enhancing patch-level semantics and enabling MLP-Mixer to better model multi-scale dependencies. MPMC further stacks RAM and MLP-Mixer in a memory-efficient manner, achieving more effective temporal mixing. It addresses semantic gaps across scales and facilitates better multiscale modeling and forecasting performance. We not only validate the effectiveness of SEMixer on 10 public datasets, but also on the \textit{2025 CCF AlOps Challenge} based on 21GB real wireless network data, where SEMixer achieves third place. The code is available at the link https://github.com/Meteor-Stars/SEMixer.

cs.LG↗

arXiv2026

HarmVideoBench: Benchmarking Harmful Video Understanding in Large Multimodal Models

Large vision-language models (LVLMs) have recently shown immense potential in automated content moderation, sparking growing interest in developing harmful-video benchmarks. However, we identify two primary limitations in existing works: 1) The multi-layered characteristics of harmful videos are overlooked. Existing benchmarks predominantly formulate evaluation as a binary classification task, failing to capture implicit or deep contextual harms. 2) Explanatory rationales are completely absent. Current frameworks measure exclusively whether a model flags a video correctly rather than explaining why, turning evaluation into a black box where models can succeed through superficial shortcuts. To address these problems, we present HarmVideoBench, a multi-layered diagnostic benchmark comprising 1,379 videos paired with 4,137 multiple-choice questions. HarmVideoBench benchmarks three hierarchical dimensions: Observable Evidence, Clip-Internal Meaning, and Beyond-Clip Reasoning, aiming to evaluate models' deep understanding beyond surface cues with carefully balanced and curated samples. We evaluate 19 leading models on HarmVideoBench to assess their multidimensional understanding of harmful videos. Moreover, we introduce BCR, a benchmark-aligned method that predicts reasoning boundaries and dynamically retrieves context only when needed. Experimental results show that BCR substantially improves the base model's performance in harmful video understanding, raising the macro average from 61.7 percent to a state-of-the-art 84.4 percent.

cs.CV↗

arXiv2026

Self-Adaptive Scale Handling for Forecasting Time Series with Scale Heterogeneity

Current time series forecasting (TSF) research predominantly focuses on scale-homogeneous data, where different time series share similar numerical magnitude ranges. However, in real-world industrial scenarios such as financial product sales, different time series often differ by orders of magnitude (scale heterogeneity). Since these series share similar temporal patterns, joint modeling is desirable for better data utilization, yet existing scaling methods either compress low-scale signals (global normalization) or destroy semantic discriminability and amplify inverse-scaling errors (window-based scaling). This paper proposes a self-Adaptive Scale-handling (AS) module that learns adaptive scale factors tailored to each input, preserving semantic discriminability while reducing inverse-scaling errors. AS consists of Scale Calibrating (SC), which calibrates prior mean scaling factors through neural networks, and Scaling Selection (SS), which decides whether to apply calibration or retain the original factor, avoiding over-calibration. Experiments on real-world fund sales datasets from Ant Fortune and Alipay show that AS seamlessly integrates into popular TSF models and consistently improves their performance. The code and dataset are available at the link https://github.com/Meteor-Stars/ASTSF.

cs.LG↗

arXiv2026

Interpretable Audio Editing Evaluation via Chain-of-Thought Difference-Commonality Reasoning with Multimodal LLMs

Automatic mean opinion score (MOS) prediction serves as a principled alternative to both subjective listening tests and objective metrics, providing scalable and consistent audio evaluation. Inspired by the LLM-as-Judge paradigm, recent multimodal large language models offer strong perceptual modeling and reasoning capabilities, enabling audio quality assessment. In this work, we address the challenging problem of audio editing evaluation and propose the first natural language-based automated evaluation framework built upon Qwen2-Audio. Two caption-based fine-tuning tasks are introduced to enhance multi-audio understanding, together with a designed Chain-of-Thought prompting strategy to encourage structured, step-by-step reasoning. Experiments show that our framework produces interpretable and logically consistent text-based evaluations, aligning closely with human judgments while outperforming existing baselines. The code and demo are available at https://github.com/NKU-HLT/Eval_Reasoning.

cs.SD↗

arXiv2026

Threshold cusp effects to measure masses of radiatively decaying hadrons: The $B_{s0}^*$ mass from the $Υϕ$ spectrum

Hadrons that decay predominantly into final states containing photons are notoriously difficult to detect at hadron colliders. Prominent examples are the yet-unobserved $B_{s0}^*$ and $B_{s1}$, the bottom partners of the $D_{s0}^*(2317)$ and $D_{s1}(2460)$, which are expected to exhibit exotic properties deviating from the conventional quark-model predictions of $\bar b s$ mesons. We propose a general, model-independent method to overcome this problem: when the target hadron has an attractive $S$-wave interaction with a companion hadron of precisely known mass, the line shape of a suitable final state develops a cusp at the pair threshold, or a peak just below it if the attraction binds, so that subtracting the companion mass returns the target mass, up to the binding energy in the latter case. As a proof of concept, a leading-order particle-dimer calculation of the $D\bar D_s K$ three-body system reproduces the $X(4274)$ structure in the LHCb $J/ψϕ$ distribution extracted from $B\to K J/ψϕ$, as a $D_{s0}^*\bar D_s$ threshold cusp driven by a nearby virtual-state pole, yielding $m_{D_{s0}^*}=(2322\pm6)$ MeV in agreement with its measured value and favoring $J^{PC}=0^{-+}$ for the $X(4274)$. Transferring the elastic three-body interaction to the bottom sector, we predict an analogous structure at the $B_{s0}^*\bar B_s$ threshold near $11.09$ GeV, making the $Υϕ$ invariant-mass distribution at the LHC a clean probe of the $B_{s0}^*$ mass.

hep-ph↗

arXiv2026

Numerical Integration at the Intersection of Probability Theory, Number Theory and Dynamical Systems

This paper is devoted to the study of uniformly distributed (u.d.) sequences on general non-compact topological spaces. Convergence-determining classes and the equivalent characterizations of u.d. sequences on $T_4$ spaces are discussed. For Polish spaces endowed with regular probability measures, the existence of a countable convergence-determining class for u.d. sequences is proved by virtue of the tightness of Borel probability measures. It is shown that, under the condition that a countable convergence-determining class exists, both independent identically distributed (i.i.d.) randomly sampled sequences and orbit sequences of ergodic transformations are u.d.\ with full measure.

math.DS↗

arXiv2026

Moving Beyond Diffusion: Hierarchy-to-Hierarchy Autoregression for fMRI-to-Image Reconstruction

Reconstructing visual stimuli from fMRI signals is a central challenge bridging machine learning and neuroscience. Recent diffusion-based methods typically map fMRI activity to a single neural embedding, using it as static guidance throughout the entire generation process. However, this fixed guidance collapses hierarchical neural information and is misaligned with the stage-dependent demands of image reconstruction. In response, we propose MindHier, a coarse-to-fine fMRI-to-image reconstruction framework built on scale-wise autoregressive modeling. MindHier introduces three components: a Hierarchical fMRI Encoder to extract multi-level neural embeddings, a Hierarchy-to-Hierarchy Alignment scheme to enforce layer-wise correspondence with CLIP features, and a Scale-Aware Coarse-to-Fine Neural Guidance strategy to inject these embeddings into autoregression at matching scales. These designs make MindHier an efficient and cognitively aligned alternative to diffusion-based methods by enabling a hierarchical reconstruction process that synthesizes global semantics before refining local details, akin to human visual perception. Extensive experiments on the NSD dataset show that MindHier achieves superior semantic fidelity, 4.67$\times$ faster inference, and more deterministic results than the diffusion-based baselines.

cs.CV↗

arXiv2026

X-ray variability of SDSS J000532.84+200717.4: from a normal state to an X-weak state

We present a multi-epoch study of the extreme X-ray variability of the type~1 quasar SDSS~J000532.84+200717.4 using archival observations from \textit{XMM-Newton}, \textit{Swift}/XRT, \textit{EP-FXT}, and \textit{ROSAT}, together with new optical spectroscopy and multi-wavelength photometry. The 0.2--10~keV X-ray flux exhibits a transition from a high state to a subsequent low state, declining by more than an order of magnitude and placing the source in the X-ray--weak regime ($Δα_{\rm ox} \lesssim -0.3$). Significant variability on timescales of days to weeks persists within the low state. In contrast, the optical and mid-infrared emission remain stable over decade-long timescales, while the UV continuum varies only mildly and broadly tracks the X-ray evolution. Multi-epoch optical spectroscopy shows no significant long-term changes in either the continuum shape or the broad emission-line profiles. The \ion{Mg}{2} emission is relatively weak compared with typical quasars, suggesting similarities to weak-line quasars. The pronounced wavelength-dependent variability indicates that the accretion disk remains largely intact while the X-ray emission undergoes dramatic changes. The spectral hardening in the low state and the viability of ionized partial-covering models are consistent with variable, largely dust-free absorbing gas, possibly associated with clumpy inner disk winds, although intrinsic coronal variations cannot be excluded. SDSS~J0005+200717.4 therefore provides evidence that extreme X-ray weakness can arise as a transient phase in otherwise normal quasars.

astro-ph.GA↗

arXiv2026

FalconGEMM: Surpassing Hardware Peaks with Lower-Complexity Matrix Multiplication

Peak breaking Matrix Multiplication is a promising technique to improve the performance of DL, especially in LLM training and inference. We present FalconGEMM, a cross-platform framework that automates the deployment, optimization, and selection of Lower-Complexity Matrix Multiplication Algorithms (LCMAs) across diverse hardware. There are three key innovations: (1) a Deployment Module that enables portable execution across various hardware and input configurations through code generation; (2) an Execution Module with Group-Parallel Optimizations that maximizes on-chip data reuse, utilizes parallel resources, and reduces bandwidth overhead; and (3) a Decision Module featuring a lightweight analytical performance model to select the optimal strategy based on matrix shapes and hardware profiles. Extensive evaluation is conducted on LLM workloads across GPU (H20, A100) and CPU (ARM, x86) architectures with multiple data types. FalconGEMM succeeds in delivering peak breaking performance and outperforms GEMM libraries (e.g., cuBLAS, CUTLASS, Intel MKL, etc) by 7.59%-17.85% and LCMA competitors like AlphaTensor by 12.41%-55.61%. Our framework makes the theoretical promise of LCMAs practical for production deployment across the heterogeneous landscape of modern hardware.

cs.DC↗

arXiv2026

SpecLoR: Spectral Lookahead Rectification for Motion-Coherent Text-to-Video Generation

Flow Matching has enabled robust text-to-video generation via latent ODE sampling. However, velocity approximation and numerical discretization errors inevitably accumulate, causing sampling trajectories to drift. Consequently, generated videos often suffer from severe spatiotemporal inconsistencies. Nevertheless, directly correcting these drifted, noisy latents is challenging: (i) timestep-dependent noise obscures reliable structural cues; (ii) spatial interventions risk disrupting intricate local geometry while incurring heavy computational costs. To address this, we propose Spectral Lookahead Rectification (SpecLoR), a plug-and-play inference method that bypasses noise via lookahead prediction, and circumvents spatiotemporal entanglement by shifting corrections to the frequency domain, where universal statistical priors of natural videos are readily available. First, during early sampling stages, SpecLoR looks ahead to estimate the clean latent $z_{t,0}$ and computes its 3D spatiotemporal spectrum. Next, SpecLoR rectifies the amplitude spectrum to match the prior, leaving the phase intact. Finally, the corrected state is re-noised to resume ODE integration. Experiments on Wan2.2 demonstrate that SpecLoR significantly reduces physical artifacts and enhances motion coherence across multiple benchmarks with minimal computational overhead (4 additional NFEs).

cs.CV↗

arXiv2026

Amortized Predictability-aware Training Framework for Time Series Forecasting and Classification

Time series data are prone to noise in various domains, and training samples may contain low-predictability patterns that deviate from the normal data distribution, leading to training instability or convergence to poor local minima. Therefore, mitigating the adverse effects of low-predictability samples is crucial for time series analysis tasks such as time series forecasting (TSF) and time series classification (TSC). While many deep learning models have achieved promising performance, few consider how to identify and penalize low-predictability samples to improve model performance from the training perspective. To fill this gap, we propose a general Amortized Predictability-aware Training Framework (APTF) for both TSF and TSC. APTF introduces two key designs that enable the model to focus on high-predictability samples while still learning appropriately from low-predictability ones: (i) a Hierarchical Predictability-aware Loss (HPL) that dynamically identifies low-predictability samples and progressively expands their loss penalty as training evolves, and (ii) an amortization model that mitigates predictability estimation errors caused by model bias, further enhancing HPL's effectiveness. The code is available at https://github.com/Meteor-Stars/APTF.

cs.LG↗

arXiv2026

Lost in the Non-convex Loss Landscape: How to Fine-tune the Large Time Series Model?

Recently, large time series models (LTSMs) have gained increasing attention due to their similarities to large language models, including flexible context length, scalability, and task generality, outperforming advanced task-specific models. However, prior studies indicate that pre-trained LTSMs may exhibit a poorly conditioned non-convex loss landscape, leading to limited trainability. As a result, direct fine-tuning tends to cause overfitting and suboptimal performance, sometimes even worse than training from scratch, substantially diminishing the benefits of pre-training. To overcome this limitation, we propose Smoothed Full Fine-tuning (SFF), a novel fine-tuning technology. Specifically, we construct an auxiliary LTSM via random initialization to obtain a smoother loss landscape, and then linearly interpolate its weights with those of the pre-trained model to smooth the original landscape. This process improves trainability while preserving pre-trained knowledge, thereby enabling more effective downstream fine-tuning. From an optimization perspective, SFF perturbs sharp minima without significantly harming flat regions, facilitating escape from poor local basins toward smoother and more generalizable solutions. Extensive experiments on benchmark datasets demonstrate consistent improvements across eight representative LTSMs, including Timer, TimesFM, MOMENT, UniTS, MOIRAI, Chronos, TTMs, and Sundial, on diverse downstream tasks. The code is available at the link: https://github.com/Meteor-Stars/SFF.

cs.LG↗