arXiv subjects

Yi Xu

Publications and source records attributed to Yi Xu.

At least 91 records · Page 5Linked to original sources

arXiv2026

Out-of-Sight Embodied Agents: Multimodal Tracking, Sensor Fusion, and Trajectory Forecasting

Trajectory prediction is a fundamental problem in computer vision, vision-language-action models, world models, and autonomous systems, with broad impact on autonomous driving, robotics, and surveillance. However, most existing methods assume complete and clean observations, and therefore do not adequately handle out-of-sight agents or noisy sensing signals caused by limited camera coverage, occlusions, and the absence of ground-truth denoised trajectories. These challenges raise safety concerns and reduce robustness in real-world deployment. In this extended study, we introduce major improvements to Out-of-Sight Trajectory (OST), a task for predicting noise-free visual trajectories of out-of-sight objects from noisy sensor observations. Building on our prior work, we expand Out-of-Sight Trajectory Prediction (OOSTraj) from pedestrians to both pedestrians and vehicles, increasing its relevance to autonomous driving, robotics, and surveillance. Our improved Vision-Positioning Denoising Module exploits camera calibration to establish vision-position correspondence, mitigating the lack of direct visual cues and enabling effective unsupervised denoising of noisy sensor signals. Extensive experiments on the Vi-Fi and JRDB datasets show that our method achieves state-of-the-art results for both trajectory denoising and trajectory prediction, with clear gains over prior baselines. We also compare with classical denoising methods, including Kalman filtering, and adapt recent trajectory prediction models to this setting, establishing a stronger benchmark. To the best of our knowledge, this is the first work to use vision-positioning projection to denoise noisy sensor trajectories of out-of-sight agents, opening new directions for future research.

cs.CV↗

arXiv2026

VehicleMemBench: An Executable Benchmark for Multi-User Long-Term Memory in In-Vehicle Agents

With the growing demand for intelligent in-vehicle experiences, vehicle-based agents are evolving from simple assistants to long-term companions. This evolution requires agents to continuously model multi-user preferences and make reliable decisions in the face of inter-user preference conflicts and changing habits over time. However, existing benchmarks are largely limited to single-user, static question-answer settings, failing to capture the temporal evolution of preferences and the multi-user, tool-interactive nature of real vehicle environments. To address this gap, we introduce VehicleMemBench, a multi-user long-context memory benchmark built on an executable in-vehicle simulation environment. The benchmark evaluates tool use and memory by comparing the post-action environment state with a predefined target state, enabling objective and reproducible evaluation without LLM-based or human scoring. VehicleMemBench includes 23 tool modules, and each sample contains over 80 historical memory events. Experiments show that powerful models perform well on direct instruction tasks but struggle in scenarios involving memory evolution, particularly when user preferences change dynamically. Even advanced memory systems struggle to handle domain-specific memory requirements in this environment. These findings highlight the need for more robust and specialized memory management mechanisms to support long-term adaptive decision-making in real-world in-vehicle systems. To facilitate future research, we release the data and code.

cs.AI↗

arXiv2026

Den-TP: A Density-Balanced Data Curation and Evaluation Framework for Trajectory Prediction

Trajectory prediction in autonomous driving has traditionally been studied from a model-centric perspective. However, existing datasets exhibit a strong long-tail distribution in scenario density, where common low-density cases dominate and safety-critical high-density cases are severely underrepresented. This imbalance limits model robustness and hides failure modes when standard evaluations average errors across all scenarios. We revisit trajectory prediction from a data-centric perspective and present Den-TP, a framework for density-aware dataset curation and evaluation. Den-TP first partitions data into density-conditioned regions using agent count as a dataset-agnostic proxy for interaction complexity. It then applies a gradient-based submodular selection objective to choose representative samples within each region while explicitly rebalancing across densities. The resulting subset reduces the dataset size by 50\% yet preserves overall performance and significantly improves robustness in high-density scenarios. We further introduce density-conditioned evaluation protocols that reveal long-tail failure modes overlooked by conventional metrics. Experiments on Argoverse 1 and 2 with state-of-the-art models show that robust trajectory prediction depends not only on data scale, but also on balancing scenario density.

cs.LG↗

arXiv2026

Null-Space Filtering for Data-Free Continual Model Merging: Preserving Stability, Promoting Plasticity

Data-free continual model merging (DFCMM) aims to fuse independently fine-tuned models into a single backbone that evolves with incoming tasks without accessing task data. This paper revisits two fundamental desiderata for DFCMM: stability, avoiding interference with earlier tasks, and plasticity, adapting faithfully to each new task. This poses a challenge that existing approaches fail to address: how to bridge data-level desiderata with parameter-space optimization to ensure stability and plasticity in the absence of task data. To this end, we propose NUFILT (NUll-space FILTering), a data-free framework that directly links these desiderata into parameter-space optimization. Our key observation is that task vectors approximately align with representation subspaces, providing structural surrogates for enforcing stability and plasticity. Accordingly, we design a null-space projector that preserves prior responses by filtering overlapping components of new task vectors, ensuring stability. We further introduce a lightweight LoRA adapter that injects complementary task-specific signals to enable plasticity. The adapter is trained with a projection-based surrogate loss that preserves consistency with prior knowledge while introducing novel directions. This joint filtering-adaptation process enables the backbone to absorb new knowledge while retaining existing behaviors, with updates fused back in a layer-wise linear fashion without extra parameters or inference cost. Theoretically, we establish approximate subspace alignment guarantees that justify null-space filtering. Empirically, NUFILT achieves state-of-the-art performance with minimal forgetting on both vision and NLP benchmarks, improving average accuracy by 4-7% over OPCM and WUDI-Merging, while narrowing the gap to fine-tuning and reducing computation overhead. The code is available at: https://github.com/zihuanqiu/NUFILT

cs.LG↗

arXiv2026

Bridging Sequential and Contextual Features with a Dual-View of Fine-grained Core-Behaviors and Global Interest-Distribution

Click-through rate (CTR) prediction tasks typically estimate the probability of a user clicking on a candidate item by modeling both user behavior sequence features and the item's contextual features, where the user behavior sequence is particularly critical as it dynamically reflects real-time shifts in user interest. Traditional CTR models often aggregate this dynamic sequence into a single vector before interacting it with contextual features. This approach, however, not only leads to behavior information loss during aggregation but also severely limits the model's capacity to capture interactions between contextual features and specific user behaviors, ultimately impairing its ability to capture fine-grained behavioral details and hindering models' prediction accuracy. Conversely, a naive approach of directly interacting with each user action with contextual features is computationally expensive and introduces significant noise from behaviors irrelevant to the candidate item. This noise tends to overwhelm the valuable signals arising from interactions involving more behaviors relevant to the candidate item. Therefore, to resolve the above issue, we propose a Core-Behaviors and Distributional-Compensation Dual-View Interaction Network (CDNet), which bridges the gap between sequential and contextual feature interactions from two complementary angles: a fine-grained interaction involving the most relevant behaviors and contextual features, and a coarse-grained interaction that models the user's overall interest distribution against the contextual features. By simultaneously capturing important behavioral details without forgoing the holistic user interest, CDNet effectively models the interplay between sequential and contextual features without imposing a significant computational burden. Ultimately, extensive experiments validate the effectiveness of CDNet.

cs.IR↗

arXiv2026

Deferred is Better: A Framework for Multi-Granularity Deferred Interaction of Heterogeneous Features

Click-through rate (CTR) prediction models estimates the probability of a user-item click by modeling interactions across a vast feature space. A fundamental yet often overlooked challenge is the inherent heterogeneity of these features: their sparsity and information content vary dramatically. For instance, categorical features like item IDs are extremely sparse, whereas numerical features like item price are relatively dense. Prevailing CTR models have largely ignored this heterogeneity, employing a uniform feature interaction strategy that inputs all features into the interaction layers simultaneously. This approach is suboptimal, as the premature introduction of low-information features can inject significant noise and mask the signals from information-rich features, which leads to model collapse and hinders the learning of robust representations. To address the above challenge, we propose a Multi-Granularity Information-Aware Deferred Interaction Network (MGDIN), which adaptively defers the introduction of features into the feature interaction process. MGDIN's core mechanism operates in two stages: First, it employs a multi-granularity feature grouping strategy to partition the raw features into distinct groups with more homogeneous information density in different granularities, thereby mitigating the effects of extreme individual feature sparsity and enabling the model to capture feature interactions from diverse perspectives. Second, a delayed interaction mechanism is implemented through a hierarchical masking strategy, which governs when and how each group participates by masking low-information groups in the early layers and progressively unmasking them as the network deepens. This deferred introduction allows the model to establish a robust understanding based on high-information features before gradually incorporating sparser information from other groups...

cs.IR↗

arXiv2026

An Ultra-Faint, Chemically Primitive Galaxy Forming in the Reionization Era

The formation of the first stars and galaxies marked the onset of chemical enrichment, yet direct observations of such primordial systems remain elusive. Here we present James Webb Space Telescope spectroscopic observations of LAP1-B, an ultra-faint galaxy at redshift z_{spec}=6.625 +/-0.001, corresponding to a cosmic age of 800 million years after the Big Bang, strongly magnified by gravitational lensing. LAP1-B exhibits a gas-phase oxygen abundance of (4.2 +/- 1.8) x 10^{-3} times the solar value, making it the most chemically primitive star-forming galaxy discovered to date. The galaxy displays an exceptionally hard ionizing radiation field, which is inconsistent with chemically enriched stellar populations or accreting black holes but matches theoretical predictions for an exceptionally metal-deficient stellar population. It also shows an elevated carbon-to-oxygen abundance ratio for its metallicity in the interstellar medium, consistent with nucleosynthetic yields from a stellar population formed in the absence of initial metals. The lack of detectable stellar continuum constrains the stellar mass to <3,300 Msun, while the dynamical mass, derived from emission-line kinematics, exceeds the combined stellar and gas mass and indicates a dominant dark matter halo. Our findings establish LAP1-B as a "fossil in the making", a direct high-redshift progenitor of the ancient ultra-faint dwarf galaxies observed in the local Universe, offering a rare window into the earliest stages of galaxy formation.

astro-ph.GA↗

arXiv2026

Taming the Long Tail: Denoising Collaborative Information for Robust Semantic ID Generation

Item IDs form the backbone of industrial recommender systems, but suffer from representation instability and poor long-tail generalization in large, dynamic item corpora. Semantic IDs (SIDs) mitigate these issues by enabling knowledge sharing through quantization of item content features. Existing methods attempt to enhance SID expressiveness by incorporating collaborative information with content features; however, they often overlook a critical distinction: unlike relatively uniform content features, user-item interactions are highly skewed, resulting in a significant quality gap in collaborative information between popular and long-tail items. This mismatch leads to two critical limitations: (1) Collaborative Noise Corrupts Behavior-Content Alignment: Behavior-content alignment is a prevailing approach for modeling shared information. However, indiscriminate alignment allows collaborative noise from long-tail items to corrupt their content representations, leading to the loss of critical multimodal information. (2) Collaborative Noise Obscures Critical Behavioral SIDs: When modeling modality-specific information, prior works typically generate multiple behavioral SIDs with equal weights for each item. This equal-weight scheme fails to reflect the varying importance of different behavioral SIDs, making it difficult for downstream tasks to distinguish informative SIDs from noisy ones. To address these challenges, we propose ADC-SID, a framework that Adaptively Denoises Collaborative information for SID quantization. It comprises two key components: (i) Adaptive Behavior-Content Alignment, which adjusts alignment strength to mitigate corruption caused by collaborative noise; and (ii) Dynamic Behavioral Weighting Mechanism, which learns importance scores for behavioral SIDs to enable downstream models to suppress noise. Extensive experiments has demonstrated ADC-SID's superiority...

cs.IR↗

arXiv2026

Pretraining with Token-Level Adaptive Latent Chain-of-Thought

Scaling large language models by increasing parameters and training data is increasingly constrained by limited high-quality corpora and rising communication costs. This work explores an alternative axis: increasing per-token computation without expanding parameters, by internalizing latent Chain-of-Thought (CoT) into pretraining. We propose Pretraining with Token-Level Adaptive Latent CoT (adaptive latent CoT), where the model generates a variable-length latent CoT trajectory before emitting each token -- allocating longer trajectories to difficult tokens and shorter (or even zero) trajectories to easy ones. Importantly, this behavior emerges naturally from one-stage pretraining on general text and reduces computation in both training and inference via token-wise adaptive halting. Experiments with Llama architectures show that adaptive latent CoT consistently improves language modeling perplexity and broad downstream accuracy, even with fewer training FLOPs than prior recurrent baselines.

cs.CL↗

arXiv2026

End-to-End Direction-Aware Keyword Spotting with Spatial Priors in Noisy Environments

Keyword spotting (KWS) is crucial for many speech-driven applications, but robust KWS in noisy environments remains challenging. Conventional systems often rely on single-channel inputs and a cascaded pipeline separating front-end enhancement from KWS. This precludes joint optimization, inherently limiting performance. We present an end-to-end multi-channel KWS framework that exploits spatial cues to improve noise robustness. A spatial encoder learns inter-channel features, while a spatial embedding injects directional priors; the fused representation is processed by a streaming backbone. Experiments in simulated noisy conditions across multiple signal-to-noise ratios (SNRs) show that spatial modeling and directional priors each yield clear gains over baselines, with their combination achieving the best results. These findings validate end-to-end multi-channel spatial modeling, indicating strong potential for the target-speaker-aware detection in complex acoustic scenarios.

eess.AS↗

arXiv2026

Nuclear Physics of X-ray Bursts

Thermonuclear X-ray bursts from the surface of accreting neutron stars are the most common astrophysical explosions in our galaxy. They provide a unique window into the physics of neutron stars, the physics of matter under extreme conditions, and the physics of astrophysical thermonuclear explosions. X-ray bursts are powered by a broad range of nuclear reactions that need to be understood to interpret observations. The relevant nuclei are mostly neutron deficient and unstable, and thus experimental information and theoretical understanding is limited and an active area of research in nuclear science. We review the current status of the nuclear physics of X-ray bursts, with special emphasis on new experimental and theoretical information on a large number of reaction rates. As such we provide an overview of the broad experimental and theoretical methods currently used to advance the nuclear physics of X-ray bursts. The new information is used to update the public JINA REACLIB database with 32 new reaction rates based on experimental information, and a new dataset of theoretical statistical model reaction rates where no experimental information is available. Using several models for X-ray bursts that are powered by mixed hydrogen and helium burning, we take advantage of the updated nuclear data to review the current understanding of the nuclear reaction sequences in such X-ray bursts, the modeling of light curves, and predictions of the composition of nuclear ashes.

astro-ph.HE↗

arXiv2026

Narrowing the Gap: SOS Ranks of $4 \times 3$ Biquadratic Forms and a Lower Bound of $8$

We investigate the maximum sum-of-squares (SOS) rank of biquadratic forms in the critical case of $4 \times 3$ variables, where the general bounds are currently $7 \leq \mathrm{BSR}(4,3) \leq 11$. By analyzing two important structured subclasses, we obtain exact determinations and improved upper bounds that significantly narrow this gap. For simple biquadratic forms those containing only distinct terms of the type $x_i^2 y_j^2$ we prove that the maximum achievable SOS rank is exactly 7, a value attained by a form corresponding to a $C_4$-free bipartite graph with the maximum number of edges. This settles the question for simple forms. For $y$-deficient biquadratic forms a class introduced here that permits cross terms among two of the three $y$-variables while the third appears only in pure square terms we prove an upper bound of $9$ by combining Calderön's theorem on $m\times 2$ forms with the known value $\mathrm{BSR}(4,2) = 5$. Our main result is a constructive proof that $\mathrm{BSR}(4,3) \geq 8$. We present an explicit non-simple, non-deficient $4\times 3$ biquadratic form and prove it requires exactly eight squares, thereby improving the general lower bound. This shows that any form achieving a rank higher than $8$ must possess a more complex algebraic structure, and it reduces the search space for determining the true value of $\mathrm{BSR}(4,3)$. Connections to Zarankiewicz numbers, extremal graph theory, and classical results on sums of squares are highlighted throughout.

math.OC↗

arXiv2026

Sum of Squares Rank of Biquadratic Forms and The Zarankiewicz Number

Denote the maximum sos rank of $m \times n$ sum of squares (SOS) biquadratic forms by $BSR(m, n)$. In this paper, we show that $BSR(m, n) \ge z(m, n)$ and conjecture that $BSR(m, n) = z(m, n)$, where $z(m, n)$ is the Zarankiewicz number. Our result coincides with the existing results for $m = 2$, $n = 2$, and $m = n = 3$, and is superior to other previously known lower bounds. Our result also connects graph theory and SOS polynomial theory.

math.OC↗

arXiv2026

Visual Planning: Let's Think Only with Images

Recent advancements in Large Language Models (LLMs) and their multimodal extensions (MLLMs) have substantially enhanced machine reasoning across diverse tasks. However, these models predominantly rely on pure text as the medium for both expressing and structuring reasoning, even when visual information is present. In this work, we argue that language may not always be the most natural or effective modality for reasoning, particularly in tasks involving spatial and geometrical information. Motivated by this, we propose a new paradigm, Visual Planning, which enables planning through purely visual representations for these "vision-first" tasks, as a supplementary channel to language-based reasoning. In this paradigm, planning is executed via sequences of images that encode step-by-step inference in the visual domain, akin to how humans sketch or visualize future actions. We introduce a novel reinforcement learning framework, Visual Planning via Reinforcement Learning (VPRL), empowered by GRPO for post-training large vision models, leading to substantial improvements in planning in a selection of representative visual navigation tasks, FrozenLake, Maze, and MiniBehavior. Our visual planning paradigm outperforms all other planning variants that conduct reasoning in the text-only space. Our results establish Visual Planning as a viable and promising supplement to language-based reasoning, opening new avenues for tasks that benefit from intuitive, image-based inference.

cs.LG↗

arXiv2026

Controlled Self-Evolution for Algorithmic Code Optimization

Self-evolution methods enhance code generation through iterative "generate-verify-refine" cycles, yet existing approaches suffer from low exploration efficiency, failing to discover solutions with superior complexity within limited budgets. This inefficiency stems from initialization bias trapping evolution in poor solution regions, uncontrolled stochastic operations lacking feedback guidance, and insufficient experience utilization across tasks. To address these bottlenecks, we propose Controlled Self-Evolution (CSE), which consists of three key components. Diversified Planning Initialization generates structurally distinct algorithmic strategies for broad solution space coverage. Genetic Evolution replaces stochastic operations with feedback-guided mechanisms, enabling targeted mutation and compositional crossover. Hierarchical Evolution Memory captures both successful and failed experiences at inter-task and intra-task levels. Experiments on EffiBench-X demonstrate that CSE consistently outperforms all baselines across various LLM backbones. Furthermore, CSE achieves higher efficiency from early generations and maintains continuous improvement throughout evolution. Our code is publicly available at https://github.com/QuantaAlpha/EvoControl.

cs.CL↗

arXiv2026

SHIELD: Suppressing Hallucinations In LVLM Encoders via Bias and Vulnerability Defense

Large Vision-Language Models (LVLMs) excel in diverse cross-modal tasks. However, object hallucination, where models produce plausible but inaccurate object descriptions, remains a significant challenge. In contrast to previous work focusing on LLM components, this paper is the first to trace LVLM hallucinations to visual encoders and identifies three key issues: statistical bias, inherent bias, and vulnerability. To address these challenges, we propose SHIELD, a training-free framework that mitigates hallucinations through three strategies: re-weighting visual tokens to reduce statistical bias, introducing noise-derived tokens to counter inherent bias, and applying adversarial attacks with contrastive decoding to address vulnerability. Experiments demonstrate that SHIELD effectively mitigates object hallucinations across diverse benchmarks and LVLM families. Moreover, SHIELD achieves strong performance on the general LVLM benchmark, highlighting its broad applicability. Code is available at https://github.com/hukcc/SHIELD.

cs.CV↗

arXiv2026

REG4Rec: Reasoning-Enhanced Generative Model for Large-Scale Recommendation Systems

Sequential recommendation aims to predict a user's next action in large-scale recommender systems. While traditional methods often suffer from insufficient information interaction, recent generative recommendation models partially address this issue by directly generating item predictions. To better capture user intents, recent studies have introduced a reasoning process into generative recommendation, significantly improving recommendation performance. However, these approaches are constrained by the singularity of item semantic representations, facing challenges such as limited diversity in reasoning pathways and insufficient reliability in the reasoning process. To tackle these issues, we introduce REG4Rec, a reasoning-enhanced generative model that constructs multiple dynamic semantic reasoning paths alongside a self-reflection process, ensuring high-confidence recommendations. Specifically, REG4Rec utilizes an MoE-based parallel quantization codebook (MPQ) to generate multiple unordered semantic tokens for each item, thereby constructing a larger-scale diverse reasoning space. Furthermore, to enhance the reliability of reasoning, we propose a training reasoning enhancement stage, which includes Preference Alignment for Reasoning (PARS) and a Multi-Step Reward Augmentation (MSRA) strategy. PARS uses reward functions tailored for recommendation to enhance reasoning and reflection, while MSRA introduces future multi-step actions to improve overall generalization. During inference, Consistency-Oriented Self-Reflection for Pruning (CORP) is proposed to discard inconsistent reasoning paths, preventing the propagation of erroneous reasoning. Lastly, we develop an efficient offline training strategy for large-scale recommendation. Experiments on real-world datasets and online evaluations show that REG4Rec delivers outstanding performance and substantial practical value.

cs.IR↗

arXiv2026

On the SOS Rank of Simple and Diagonal Biquadratic Forms

We study the sum-of-squares (SOS) rank of simple and diagonal biquadratic forms. For simple biquadratic forms in $3 \times 3$ variables, we show that the maximum SOS rank is exactly $6$, attained by a specific six-term form. We further prove that for any $m \ge 3$, there exists an $m \times m$ simple biquadratic form whose SOS rank is exactly $2m$. Moreover, we show that for all $m, n \ge 3$, the maximum SOS rank over $m \times n$ simple biquadratic forms is at least $m+n$, which implies $\mathrm{BSR}(m,n) \ge m+n$. For diagonal biquadratic forms with nonnegative coefficients, we prove an SOS rank upper bound of $7$, improving the general bound of $8$ for $3 \times 3$ forms. These results provide new lower and upper bounds on the worst-case SOS rank of biquadratic forms and highlight the role of structure in reducing the required number of squares.

math.OC↗