arXiv subjects

Qi Xu

Publications and source records attributed to Qi Xu.

At least 37 records · Page 2Linked to original sources

arXiv2026

A Simple Efficiency Incremental Learning Framework via Vision-Language Model with Nonlinear Multi-Adapters

Incremental Learning (IL) aims to learn new tasks while preserving previously acquired knowledge. Integrating the zero-shot learning capabilities of pre-trained vision-language models into IL methods has marked a significant advancement. However, these methods face three primary challenges: (1) the need for improved training efficiency; (2) reliance on a memory bank to store previous data; and (3) the necessity of a strong backbone to augment the model's capabilities. In this paper, we propose SimE, a Simple and Efficient framework that employs a vision-language model with adapters designed specifically for the IL task. We report a remarkable phenomenon: there is a nonlinear correlation between the number of adaptive adapter connections and the model's IL capabilities. While increasing adapter connections between transformer blocks improves model performance, adding more adaptive connections within transformer blocks during smaller incremental steps does not enhance, and may even degrade the model's IL ability. Extensive experimental results show that SimE surpasses traditional methods by 9.6% on TinyImageNet and outperforms other CLIP-based methods by 5.3% on CIFAR-100. Furthermore, we conduct a systematic study to enhance the utilization of the zero-shot capabilities of CLIP. We suggest replacing SimE's encoder with a CLIP model trained on larger datasets (e.g., LAION2B) and stronger architectures (e.g., ViT-L/14).

cs.CV↗

arXiv2026

Towards Efficient Inference under Nonmonotone Missingness with General Imputation

Missing data are ubiquitous in classical survey and longitudinal studies as well as modern multi-modality data analysis. A longstanding challenge arises under nonmonotone missingness, where different units may observe arbitrary subsets of all variables. We study parameter estimation and inference problem under this setting. Semiparametric efficiency theory characterizes the efficient estimator through inversion of an operator constructed from pattern-specific conditional expectations. However, this estimator is generally not tractable due to compositions of conditional expectations across patterns. We introduce the Restricted ANOVA hierarchY (RAY), a functional decomposition that reveals an almost-eigen structure of the operator under missing completely at random. This structure yields a closed-form, computable approximation to the efficient estimator. RAY estimator is applicable to general Z-estimation problems, and it remains unbiased for arbitrary independent imputation functions. In theory, we establish verifiable sufficient conditions where RAY attains the efficiency lower bound, and offer a general bound for the efficiency gap otherwise. We further develop adaptive RAY estimator, which attains the minimal asymptotic variance within a broader class containing RAY and other existing estimators. Finally, we investigate the extension of RAY under missing at random mechanism. Simulations and a single-cell multi-omics application demonstrate the efficiency gains of the proposed estimators.

stat.ME↗

arXiv2026

Probabilistic Residual Learning for Online Recommendations

Modern recommender systems are typically based on deep learning (DL) models, where a dense encoder learns representations of users and items. As a result, these systems often suffer from the black-box nature and computational complexity of the underlying models, making it difficult to systematically enhance their recommendation capabilities. To address this problem, we propose Probabilistic Residual Learning (PRL), a causal Bayesian recommendation model that models the residual between ground-truth and base predictions, enabling targeted refinement of existing systems. Specifically, PRL (1) probabilistically groups users for localized residual modeling, (2) models domain-level confounders that influence user and item representations, and (3) aggregates cluster-specific residual predictions over the confounders using do-calculus. Experiments demonstrate that our plug-and-play PRL is compatible with various base deep learning recommender systems, improving their performance while automatically discovering meaningful user clusters.

cs.IR↗

arXiv2026

HiCI: Hierarchical Construction-Integration for Long-Context Attention

Long-context language modeling is commonly framed as a scalability challenge of token-level attention, yet local-to-global information structuring remains largely implicit in existing approaches. Drawing on cognitive theories of discourse comprehension, we propose HiCI (Hierarchical Construction--Integration), a hierarchical attention module that constructs segment-level representations, integrates them into a shared global context, and broadcasts both to condition segment-level attention. We validate HiCI through parameter-efficient adaptation of LLaMA-2 with only <5.5% additional parameters, extending context from 4K to 100K tokens (7B) and 64K tokens (13B). Across language modeling, retrieval, and instruction-following benchmarks, HiCI yields consistent improvements over strong baselines, including matching proprietary models on topic retrieval and surpassing GPT-3.5-Turbo-16K on code comprehension. These results demonstrate the effectiveness of explicit hierarchical structuring as an inductive bias for long-context modeling.

cs.CL↗

arXiv2026

Gevrey instability in the inviscid inflow-outflow problem

We consider the 2D incompressible Euler equations on a periodic channel $\mathbb{T}\times (0,1)$ with inflow-outflow boundary condition $u=(0,1)$ on $\mathbb{T} \times \{0,1 \}$. We also impose the incoming vorticity boundary condition $ω=η$ on $\mathbb{T}\times \{ 0 \}$, where $η$ is prescribed. We show that the problem is globally well-posed in Gevrey spaces (for any value of the Gevrey exponent $s>1$) as long as $η$ remains Gevrey. This proves that the inflow-outflow velocity boundary condition determines the solution locally in time if and only if the solution is considered in an analytic class. In particular, leaving the analytic class, nonuniquness of solutions occurs already in any Gevrey class, by prescribing $η$. Hence, prescribing an analytic inflow-outflow velocity leads to precisely one analytic and continuum $s$-Gevrey solutions for every $s>1$. Furthermore, the result implies that if $η$ is analytic, then the unique global solution can lose analyticity in space for all $t>0$, but remain $s$-Gevrey regular for all $s$.

math.AP↗

arXiv2026

Towards One-to-Many Temporal Grounding

Temporal Grounding (TG) aims to localize video segments corresponding to a textual query. Prior research predominantly focuses on single-segment retrieval. Real-world scenarios, however, often require localizing multiple disjoint segments for a single query -- a setting we term One-to-Many Temporal Grounding (OMTG). Previous state-of-the-art MLLMs, optimized for one-to-one settings, struggle in this context, often yielding near-zero scores due to a lack of event cardinality perception. To bridge this gap, we present a systematic solution with three key contributions. First, we establish the first comprehensive OMTG benchmark, introducing Count Accuracy (C-Acc) and Effective Temporal F1 (EtF1) as evaluation metrics. Second, we curate a high-quality OMTG dataset comprising 56k samples through a sophisticated construction pipeline. Third, we develop novel temporal and caption reward functions specifically designed for OMTG. In particular, the caption reward leverages Chain-of-Thought reasoning over dense video captions to explicitly guide policy optimization toward both preciseness and completeness. Extensive experiments show our model achieves a new state-of-the-art EtF1 of 43.65\% on OMTG Bench, outperforming Gemini 2.5 Pro and Seed-1.8 by 15.85\% and 15.61\%, respectively. Project Page: https://insomniaaac.github.io/OMTG/

cs.CV↗

arXiv2026

Refining the Gaia DR3 Parallax Zero-point: A Hybrid Approach Combining Global Parametric Correction with Local Refinement

The Gaia Data Release 3 (GDR3) parallaxes are affected by a complex bias that depends on stellar magnitude, color, and celestial position, with amplitudes reaching tens of microarcseconds ($μ$as). Standard global parametric models (e.g., Lindegren et al. 2021, hereafter L21) effectively remove large-scale trends but struggle to resolve small-scale spatial systematics due to functional rigidity. We aim to construct a flexible, data-driven calibration map that eliminates these residual local systematics without imposing rigid functional forms. We propose a "Global Pre-correction + Local Refinement" hybrid strategy. First, we utilize the L21 model as a baseline to remove the dominant magnitude and color-dependent biases. Second, we model the residual zero-point using a Local Non-parametric method based on a Sliding Window technique. This approach fits local trends using k-nearest neighbors from quasars (for faint stars, G>18) and wide binaries combined with Large Magellanic Cloud (LMC) (for bright stars, G < 18). Our hybrid model demonstrates significant improvements over the standard L21 solution. Validation against different samples reveals a remarkably flat residual map with near-zero bias across the full sky. Our mathematical attempt at calibrating the parallax zero-point is expected to provide a useful reference for the zero-point correction in future Gaia DR4, and to help move towards a physical resolution of this issue.

astro-ph.IM↗

arXiv2026

Optical-morphology-based assessment of astrometric quality in Gaia-CRF3 quasars

Context. Several studies have shown that host-galaxy structure or extended optical morphology in AGNs can induce spurious parallaxes and proper motions in Gaia DR3. However, it remains unclear whether source morphology also introduces systematic errors into the celestial reference frame constructed from Gaia data. Aims. We aim to provide a Gaia-independent external morphological indicator for Gaia-CRF3 sources and to use it to quantify the astrometric systematics associated with source morphology. Methods. Using morphological parameters derived from DESI, SDSS, and SkyMapper, together with the PS1-PSC point-source score as a common reference scale, we used XGBoost to infer external morphological scores for Gaia-CRF3 sources. We then developed a multi-survey fusion scheme to combine the four survey-based point-source scores into a single composite score that measures the degree to which each source departs from the morphology of an ideal point source. Results. We obtained morphological scores for 1,607,490 Gaia-CRF3 sources, corresponding to a completeness of 99.59\% with respect to the full Gaia-CRF3 catalogue. The score ranges from 0 to 1 and remains reliable for sources with $G<20.85$ mag. Based on this indicator, we find that AGNs with strongly non-point-like morphology induce a parallax zero-point shift of about $-43.7\,μ$as, which cannot be effectively removed by the current parallax zero-point correction model. We also find that reference-source subsamples selected in different score ranges exhibit significantly different all-sky proper motion fields. For the high-purity point-source subsample with \texttt{point\_score} > 0.95, the total frame spin amplitude is reduced by 15.8\% relative to that of the full Gaia-CRF3 sample.

astro-ph.GA↗

arXiv2026

Watch, Remember, Reason: Human-View Video Understanding with MLLMs

Video understanding is being rapidly transformed by multimodal large language models (MLLMs), as research moves from short clips to long, multimodal, and knowledge-intensive video scenarios. These scenarios require models to handle sparse evidence, long-range dependencies, multimodal alignment, and reliable inference under limited computational budgets. This work presents a human-view perspective on LLM-based video understanding, organized around three functional abilities: watching, remembering, and reasoning. Rather than treating video tasks as isolated benchmarks, this view provides a unified structure for analyzing how video MLLMs acquire evidence, preserve context, and produce grounded outputs. We introduce a formulation that characterizes video understanding systems by their perceptual representations, memory states, reasoning traces, and final predictions. Based on this formulation, we identify challenges in spatio-temporal perception, efficient long-video processing, memory modeling, streaming understanding, and faithful reasoning. Representative methods are organized by their roles in video MLLM systems. Watching covers fine-grained, comprehensive, audio-visual, and efficient perception. Remembering includes offline and streaming memory, while reasoning covers text-only reasoning and thinking with videos. We further examine application domains such as egocentric, sports, instructional, medical, and narrative videos, and cover training datasets and evaluation benchmarks across task types, supervision formats, modalities, and capability dimensions. Finally, we outline open problems and future directions for scalable, memory-aware, and evidence-grounded video intelligence. Related works will be continuously traced at https://github.com/marinero4972/Awesome-HumanView-VideoUnderstanding.

cs.CV↗

arXiv2026

Euler Immersion

We address the Euler immersion problem, a fluid-structure interaction problem in which an elastic body is immersed in an incompressible inviscid fluid governed by the Euler equations. We show that the system exhibits a loss of one derivative and formulate the problem in analytic function spaces. We then prove local well-posedness in analytic spaces under velocity-matching boundary conditions. Finally, by means of an example, we show that existence fails in analytic spaces when both velocity and stress-matching boundary conditions are prescribed.

math.AP↗

arXiv2026

LMM-IR: Large-Scale Netlist-Aware Multimodal Framework for Static IR-Drop Prediction

Static IR drop analysis is a fundamental and critical task in the field of chip design. Nevertheless, this process can be quite time-consuming, potentially requiring several hours. Moreover, addressing IR drop violations frequently demands iterative analysis, thereby causing the computational burden. Therefore, fast and accurate IR drop prediction is vital for reducing the overall time invested in chip design. In this paper, we firstly propose a novel multimodal approach that efficiently processes SPICE files through large-scale netlist transformer (LNT). Our key innovation is representing and processing netlist topology as 3D point cloud representations, enabling efficient handling of netlist with up to hundreds of thousands to millions nodes. All types of data, including netlist files and image data, are encoded into latent space as features and fed into the model for static voltage drop prediction. This enables the integration of data from multiple modalities for complementary predictions. Experimental results demonstrate that our proposed algorithm can achieve the best F1 score and the lowest MAE among the winning teams of the ICCAD 2023 contest and the state-of-the-art algorithms.

cs.LG↗

arXiv2026

Semiparametric semi-supervised learning for general targets under distribution shift and decaying overlap

In modern scientific applications, large volumes of covariate data are readily available, while outcome labels are costly, sparse, and often subject to distribution shift. This asymmetry has spurred interest in semi-supervised (SS) learning, but most existing approaches rely on strong assumptions -- such as missing completely at random (MCAR) labeling or strict positivity -- that put substantial limitations on their practical usefulness. In this work, we introduce a general semiparametric framework for estimation, inference, and efficiency benchmarking in SS settings where labels are missing at random (MAR) and the overlap may vanish as sample size increases. Our framework, that we label D2S3, accommodates a wide range of smooth statistical targets -- including means, linear regression coefficients, quantiles, and causal effects -- and remains valid under high-dimensional nuisance estimation and distributional shift between labeled and unlabeled samples. We extend the theoretical guarantees of augmented inverse probability weighting estimators to preserve double robustness, asymptotic normality, and semiparametric efficiency under this challenging D2S3 regime. A key insight is that classical root-n convergence fails under vanishing overlap; we instead provide corrected asymptotic rates that capture the impact of the decay in overlap. We validate our theory through simulations and demonstrate practical utility in real-world applications on the internet of things and public health where labeled data are scarce.

math.ST↗

arXiv2026

Masked Generative Transformer Is What You Need for Image Editing

Diffusion models dominate image editing, yet their global denoising mechanism entangles edited regions with surrounding context, causing modifications to propagate into areas that should remain intact. We propose a fundamentally different approach by leveraging Masked Generative Transformers (MGTs), whose localized token-prediction paradigm naturally confines changes to intended regions. We present EditMGT, an MGT-based editing framework that is the first of its kind. Our approach employs multi-layer attention consolidation to aggregate cross-attention maps into precise edit localization signals, and region-hold sampling to explicitly prevent token flipping in non-target areas. To support training, we construct CrispEdit-2M, a 2M-sample high-resolution (>1024) editing dataset spanning seven categories. With only 960M parameters, EditMGT achieves state-of-the-art image similarity on multiple benchmarks while delivering 6x faster editing, demonstrating that MGTs offer a compelling alternative to diffusion-based editing.

cs.CV↗

arXiv2026

Partially-shared Imaging Regression on Integrating Heterogeneous Brain-Cognition Associations across Alzheimer's Diagnoses

Alzheimer's Disease Neuroimaging Initiative (ADNI) diagnostic groups present strong heterogeneous associations among demographic, imaging, and cognitive data. We propose a novel PArtially-shared Imaging Regression (PAIR) model to represent imaging coefficients as weighted combinations of smooth spatial components. A Total Variation penalty is applied to enforce spatial smoothness, and a Selective Integration penalty is introduced to adaptively learn partial-sharing structures across groups. Theoretically, we establish minimax-optimal error bounds that dynamically adapt to varying sharing paradigms. Numerically, PAIR achieves predictive accuracy comparable to advanced deep learning models while providing superior interpretability. Applied to ADNI data, PAIR reveals substantial heterogeneity in brain-cognition pathways between cognitively normal (CN) and cognitively impaired (CI) groups, with hippocampal imaging contributing minimally in the CN group but substantially in the CI group, particularly in the CA1, CA3, and presubiculum subfields.

stat.ME↗

arXiv2026

NTIRE 2026 3D Restoration and Reconstruction in Real-world Adverse Conditions: RealX3D Challenge Results

This paper presents a comprehensive review of the NTIRE 2026 3D Restoration and Reconstruction (3DRR) Challenge, detailing the proposed methods and results. The challenge seeks to identify robust reconstruction pipelines that are robust under real-world adverse conditions, specifically extreme low-light and smoke-degraded environments, as captured by our RealX3D benchmark. A total of 279 participants registered for the competition, of whom 33 teams submitted valid results. We thoroughly evaluate the submitted approaches against state-of-the-art baselines, revealing significant progress in 3D reconstruction under adverse conditions. Our analysis highlights shared design principles among top-performing methods and provides insights into effective strategies for handling 3D scene degradation.

cs.CV↗

arXiv2026

NTIRE 2026 Challenge on Bitstream-Corrupted Video Restoration: Methods and Results

This paper reports on the NTIRE 2026 Challenge on Bitstream-Corrupted Video Restoration (BSCVR). The challenge aims to advance research on recovering visually coherent videos from corrupted bitstreams, whose decoding often produces severe spatial-temporal artifacts and content distortion. Built upon recent progress in bitstream-corrupted video recovery, the challenge provides a common benchmark for evaluating restoration methods under realistic corruption settings. We describe the dataset, evaluation protocol, and participating methods, and summarize the final results and main technical trends. The challenge highlights the difficulty of this emerging task and provides useful insights for future research on robust video restoration under practical bitstream corruption.

cs.CV↗

arXiv2026

The Second Challenge on Cross-Domain Few-Shot Object Detection at NTIRE 2026: Methods and Results

Cross-domain few-shot object detection (CD-FSOD) remains a challenging problem for existing object detectors and few-shot learning approaches, particularly when generalizing across distinct domains. As part of NTIRE 2026, we hosted the second CD-FSOD Challenge to systematically evaluate and promote progress in detecting objects in unseen target domains under limited annotation conditions. The challenge received strong community interest, with 128 registered participants and a total of 696 submissions. Among them, 31 teams actively participated, and 19 teams submitted valid final results. Participants explored a wide range of strategies, introducing innovative methods that push the performance frontier under both open-source and closed-source tracks. This report presents a detailed overview of the NTIRE 2026 CD-FSOD Challenge, including a summary of the submitted approaches and an analysis of the final results across all participating teams. Challenge Codes: https://github.com/ohMargin/NTIRE2026_CDFSOD.

cs.CV↗

arXiv2026

NTIRE 2026 Challenge on Short-form UGC Video Restoration in the Wild with Generative Models: Datasets, Methods and Results

This paper presents an overview of the NTIRE 2026 Challenge on Short-form UGC Video Restoration in the Wild with Generative Models. This challenge utilizes a new short-form UGC (S-UGC) video restoration benchmark, termed KwaiVIR, which is contributed by USTC and Kuaishou Technology. It contains both synthetically distorted videos and real-world short-form UGC videos in the wild. For this edition, the released data include 200 synthetic training videos, 48 wild training videos, 11 validation videos, and 20 testing videos. The primary goal of this challenge is to establish a strong and practical benchmark for restoring short-form UGC videos under complex real-world degradations, especially in the emerging paradigm of generative-model-based S-UGC video restoration. This challenge has two tracks: (i) the primary track is a subjective track, where the evaluation is based on a user study; (ii) the second track is an objective track. These two tracks enable a comprehensive assessment of restoration quality. In total, 95 teams have registered for this competition. And 12 teams submitted valid final solutions and fact sheets for the testing phase. The submitted methods achieved strong performance on the KwaiVIR benchmark, demonstrating encouraging progress in short-form UGC video restoration in the wild.

cs.CV↗