Search arXiv⌕ Search

arXiv subjects

Wei Yu

Publications and source records attributed to Wei Yu.

At least 37 records · Page 2Linked to original sources
arXiv2026

YouZhi: Towards High-Concurrency Financial LLMs via Adaptive GQA-to-MLA Transition

Large language models (LLMs) drive significant financial innovations, yet their high-concurrency deployment is severely bottlenecked by KV cache memory overhead, which inflates infrastructure costs and throttles scalability. To address this, we propose YouZhi-LLM, a highly efficient financial LLM empowered by a comprehensive structural transition and training pipeline natively built on the Huawei Ascend ecosystem. At its algorithmic core, YouZhi-LLM features a layer-adaptive GQA-to-MLA transition framework that dynamically assigns per-layer FreqFold sizes, maximizing KV-cache compression while minimizing perplexity degradation. To recover representation capacity and inject domain expertise, the Ascend-based training pipeline seamlessly integrates generalized knowledge distillation with financial-specific supervised fine-tuning. Evaluations demonstrate the superiority of this systematic approach, with the adaptive transition reducing perplexity degradation by up to 35% over uniform baselines. Crucially, when evaluated on Ascend NPUs via vLLM-Ascend, the massive KV-cache reduction translates directly into deployment efficiency. Compared to their respective base models, YouZhi-7B yields a 12.3% improvement in average financial benchmark score alongside a 2.69$\times$ increase in maximum concurrency; similarly, YouZhi-14B achieves a 7.0% accuracy gain and a 2.43$\times$ concurrency boost, establishing a new paradigm for cost-effective, high-throughput financial inference.

cs.CL↗
arXiv2026

A candidate cyclotron line at 1.89 keV in the ultraluminous X-ray source NGC 4861 X-2

In this Letter, we report the detection of an absorption-like feature at ~1.89 keV in Chandra/ACIS spectra of the ultraluminous X-ray source NGC 4861 X-2, based on the deepest observation (ObsID 20992; ~58 ks). The feature is consistently recovered across independent continuum models and significantly improves the fit statistics. Monte Carlo simulations yield a detection significance of ~3.5-4.1 sigma, depending on the adopted continuum, and a blind line scan reveals a single, localized peak at the same energy. The observed properties are consistent with a proton cyclotron resonant scattering feature (CRSF), implying a magnetic field strength of B ~(3-4) x 10^14 G. The spectrum is well described by a multicolor disk blackbody (diskbb) with kTin ~0.8 keV or a strongly curved continuum with a low cutoff energy (cutoffpl; Ecut ~1.3 keV). The source shows variability confined to the soft X-ray band in the two Chandra observations where the absorption-like feature is detected. In these observations, a candidate periodic signal at P ~7.4 s is also detected, with a global significance of ~2.5 sigma.

astro-ph.HE↗
arXiv2026

The Broadband X-ray Spectral Properties during the Rising Phases of the Outburst of the New Black Hole X-ray Binary Candidate Swift J1727.8-1613

We report data analysis results about the outburst evolution and spectral properties during the hard state of the recently discovered X-ray transient Swift J1727.8--163 as observed by \emph{Insight}-HXMT. The broadband X-ray spectrum of Swift J1727.8--163 is more complex than the most typical spectral patterns of black hole X-ray binary systems, with not only a comparatively weaker reflection component but also an additional high-energy component, manifesting itself as a hard X-ray tail beyond the thermal Comptonization description detectable below 100 keV. Using reflection models combining thermal ({\tt reflkerr}) and hybrid ({\tt reflkerr\_bb}) Comptonization, we find that the emission arises from a plasma containing both thermal and non-thermal electrons, with the non-thermal tail forming a smooth extension of the thermal continuum. This supports a compact hybrid corona. We find that the inner disk radius remains at the ISCO and the reflection fraction stays low, with no abrupt spectral evolution across epochs. These results highlight the role of hybrid corona in shaping hard-state emission in black hole X-ray binaries.

astro-ph.HE↗
arXiv2026

GDformer: Going Beyond Subsequence Isolation for Multivariate Time Series Anomaly Detection

Unsupervised anomaly detection of multivariate time series is a challenging task, given the requirements of deriving a compact detection criterion without accessing the anomaly points. The existing methods are mainly based on reconstruction error or association divergence, which are both confined to isolated subsequences with limited horizons, hardly promising unified series-level criterion. In this paper, we propose the Global Dictionary-enhanced Transformer (GDformer) with a renovated dictionary-based cross attention mechanism to cultivate the global representations shared by all normal points in the entire series. Accordingly, the cross-attention maps reflect the correlation weights between the point and global representations, which naturally leads to the representation-wise similarity-based detection criterion. To foster more compact detection boundary, prototypes are introduced to capture the distribution of normal point-global correlation weights. GDformer consistently achieves state-of-the-art unsupervised anomaly detection performance on five real-world benchmark datasets. Further experiments validate the global dictionary has great transferability among various datasets.

cs.LG↗
arXiv2026

AffordVLA: Injecting Affordance Representations into Vision-Language-Action Models via Implicit Feature Alignment

Recent advances in Vision-Language-Action (VLA) models have shown strong potential for general-purpose robotic manipulation. However, the visual representations of most VLA models are often dominated by global object appearance and struggle to focus on task-relevant functional interaction regions, which limits their robustness in unstructured environments. Existing affordance-based methods typically rely on explicit mask injection or external perception modules, requiring additional annotations while introducing cascading perception errors and inference overhead. To address these limitations, we propose AffordVLA, an affordance-enhanced VLA framework that internalizes manipulation-centric affordance perception into VLA visual representations through implicit representation alignment. Specifically, we construct a zero-shot affordance teacher to extract task-conditioned affordance visual representations from RGB observations and language instructions. AffordVLA aligns the intermediate visual representations of the VLA with the affordance visual representations extracted by the teacher, thereby implicitly injecting manipulation-centric affordance perception into VLA visual representations and improving action accuracy. Extensive simulation and real-world experiments demonstrate that AffordVLA and its affordance teacher achieve state-of-the-art performance and outperform strong baselines. Ablation analyses show that AffordVLA effectively reshapes VLA visual representations while preserving inference efficiency, leading to improved manipulation success rates and training efficiency.

cs.RO↗
arXiv2026

Uplink-Downlink Duality for Beamforming in Integrated Sensing and Communications

This paper considers the beamforming and power optimization problem for a class of integrated sensing and communications (ISAC) problems that utilize the communication signals simultaneously for sensing. We formulate the problem of minimizing the Bayesian Cramér-Rao bound (BCRB) on the mean-squared error of estimating a vector of parameters, while satisfying downlink signal-to-interference-and-noise-ratio constraints for a set of communication users at the same time. The proposed optimization framework comprises two key new ingredients. First, we show that the BCRB minimization problem corresponds to maximizing beamforming power along certain sensing directions of interest. Second, the classical uplink-downlink duality for multiple-input multiple-output communications can be extended to the ISAC setting, but unlike the classical communication problem, the dual uplink problem for ISAC may entail negative noise power and needs to include an extra condition on the uplink beamformers. This new duality theory opens doors for efficient iterative algorithm for optimizing power and beamformers for ISAC.

cs.IT↗
arXiv2026

SID: Sliding into Distribution for Robust Few-Demonstration Manipulation

Generalizing robotic manipulation across object poses, viewpoints, and dynamic disturbances is difficult, especially with only a few demonstrations. End-to-end visuomotor policies are expressive but data-hungry, while planning and optimization satisfy explicit constraints but do not directly capture the interaction strategies demonstrated by humans. We propose Sliding into Distribution (SID), a structured framework that learns an object-centric motion field from canonicalized demonstrations to iteratively slide the system toward the demonstrated manifold and into the reliable operating region of a lightweight egocentric execution policy, mitigating out-of-distribution (OOD) execution. The motion field provides large corrective motions when far from the demonstration manifold and naturally vanishes near convergence, enabling robust reaching under substantial pose and viewpoint shifts. Within the reached regime, an egocentric policy trained with conditioned flow matching performs task-specific manipulation, supported by kinematically consistent point-cloud reprojection augmentation that preserves action-observation consistency. Across six real-world tasks, SID achieves approximately 90% success under OOD initializations with only two demonstrations, with under a 10% drop under distractors and external disturbances. Overall, SID provides a new paradigm for few-shot manipulation: explicitly managing distribution shift via online distribution recovery.

cs.RO↗
arXiv2026

EmambaIR: Efficient Visual State Space Model for Event-guided Image Reconstruction

Recent event-based image reconstruction methods predominantly rely on Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs) to process complementary event information. However, these architectures face fundamental limitations: CNNs often fail to capture global feature correlations, whereas ViTs incur quadratic computational complexity (e.g., $O(n^2)$), hindering their application in high-resolution scenarios. To address these bottlenecks, we introduce EmambaIR, an Efficient visual State Space Model designed for image reconstruction using spatially sparse and temporally continuous event streams. Our framework introduces two key components: the cross-modal Top-k Sparse Attention Module (TSAM) and the Gated State-Space Module (GSSM). TSAM efficiently performs pixel-level top-k sparse attention to guide cross-modal interactions, yielding rich yet sparse fusion features. Subsequently, GSSM utilizes a nonlinear gated unit to enhance the temporal representation of vanilla linear-complexity ($O(n)$) SSMs, effectively capturing global contextual dependencies without the typical computational overhead. Extensive experiments on six datasets across three diverse image reconstruction tasks - motion deblurring, deraining, and High Dynamic Range (HDR) enhancement - demonstrate that EmambaIR significantly outperforms state-of-the-art methods while offering substantial reductions in memory consumption and computational cost. The source code and data are publicly available at: https://github.com/YunhangWickert/EmambaIR

cs.CV↗
arXiv2026

Multi-Carrier Modulation: An Evolution from Time-Frequency Domain to Delay-Doppler Domain

The recently proposed orthogonal delay-Doppler division multiplexing (ODDM) modulation, which is a delay-Doppler (DD) domain multi-carrier (DDMC) modulation scheme based on the DD domain orthogonal pulse (DDOP), is studied. We first revisit the linear time-varying (LTV) channel model for the wireless channel, and review the conventional multi-carrier (MC) modulation schemes and their design guidelines for both linear time-invariant (LTI) and LTV channels. We then focus on the representation of the LTV channel in an equivalent sampled DD (ESDD) domain, and propose an impulse-function-based transmission strategy for the ESDD channel. Next, we take an in-depth look into the DDOP and show that it achieves orthogonality with respect to the fine time and frequency resolutions in the ESDD domain thus behaves like an impulse function. This allows us to unveil the unique input-output relation of the resultant ODDM modulation over the ESDD channel. We point out that the conventional MC modulation design guidelines based on the Weyl-Heisenberg (WH) frame theory can be relaxed without compromising its orthogonality or violating the WH frame theory. More specifically, for a practical communication system with bandwidth and duration constraints, MC modulation signals can be designed considering so-called local or sufficient (bi)orthogonality, which refers to the (bi)orthogonality among a WH subset for the MC signal within a specific bandwidth and duration. This novel design guideline could potentially open up opportunities for developing future waveforms required by new applications such as communication systems associated with high delay and/or Doppler shifts, as well as integrated sensing and communications.

cs.IT↗
arXiv2026

Reverberation lags viewed in hard X-rays from an accreting stellar-mass black hole

Accreting black holes are thought to swallow matter in the form of a disk and a hot cloud of plasma that glows brightly in X-rays, known as the corona. The X-ray emitting region is far too small to be directly imaged, but rapid variability of the X-ray signal can be used to infer the geometry by measuring time lags caused by material propagating towards the black hole and by coronal X-rays reflecting off the disk to imprint a reverberation lag. Reverberation lags can be recognized by characteristic spectral features, including an iron emission line at $\sim 6.4$ keV and a broad Compton hump peaking at $\sim 30$ keV. These reverberation features have both previously been detected for a few supermassive black holes in active galactic nuclei (AGNs). However, it is much more challenging to detect reverberation lags from stellar-mass black holes because they are more than a million times smaller. Previous reverberation lag measurements for stellar-mass black holes in X-ray binary systems have thus been limited to energies below 10 keV. Here we report on the first detection of the Compton hump reverberation feature from an X-ray binary, achieved by measuring lags in the broad energy range of $\sim 1-150$ keV. The accompanying detection of an iron line feature confirms the scenario of X-ray reverberation and provides strong evidence that the accretion flows in AGNs and X-ray binaries are governed by an ubiquitous process. Reverberation lags are prominent only in the most rapid variability, whereas lags in the slower variability are commonly attributed to propagating mass accretion rate perturbations. Our lag measurements up to the highest energy to date reveal that this lag in the slower variability evolves dramatically on timescales of days.

astro-ph.HE↗
arXiv2026

Site-Specific Channel Modeling and Optimization of RIS-Assisted Multiuser MISO Systems

This paper presents a physics-based channel modeling and optimization framework for reconfigurable intelligent surface (RIS)-assisted downlink multi-user multiple-input single-output (MU-MISO) communication systems in site-specific environments. A hybrid ray-tracing (RT) and full-wave electromagnetic analysis approach is developed to construct a deterministic channel model that explicitly captures multipath propagation, RIS scattering behavior, and mutual coupling effects through a non-diagonal load impedance representation. Based on this model, an alternating optimization scheme jointly updates the base-station (BS) beamformer and RIS load impedances to maximize the minimum achievable rate under a total transmit power constraint and practical capacitance limits. The objective of the proposed framework is to provide a reliable initial assessment of the system-level impact of RIS deployment in realistic propagation scenarios. To evaluate this capability, the RIS is operated in a column-paired 1-bit control mode that enables exhaustive evaluation of all realizable configurations in both simulation and measurement. Performance is compared at the distribution level through achievable-rate histograms across all configurations and further examined under small user-location variations. The observed agreement between simulation and measurement demonstrates that the proposed framework reliably captures practical performance trends and provides useful guidance for the design and deployment of RIS-assisted MU-MISO systems in site-specific environments.

eess.SP↗
arXiv2026

Refining Few-Step Text-to-Multiview Diffusion via Reinforcement Learning

Text-to-multiview (T2MV) diffusion models have shown great promise in generating multiple views of a scene from a single text prompt. While few-step backbones enable real-time T2MV generation, they often compromise key aspects of generation quality, such as per-view fidelity and cross-view consistency. Reinforcement learning (RL) finetuning offers a potential solution, yet existing approaches designed for single-image diffusion do not readily extend to the few-step T2MV setting, as they neglect cross-view coordination and suffer from weak learning signals in few-step regimes. To address this, we propose MVC-ZigAL, a tailored RL finetuning framework for few-step T2MV diffusion models. Specifically, its core insights are: (1) a new MDP formulation that jointly models all generated views and assesses their collective quality via a joint-view reward; (2) a novel advantage learning strategy that exploits the performance gains of a self-refinement sampling scheme over standard sampling, yielding stronger learning signals for effective RL finetuning; and (3) a unified RL framework that extends advantage learning with a Lagrangian dual formulation for multiview-constrained optimization, balancing single-view and joint-view objectives through adaptive primal-dual updates under a self-paced threshold curriculum that harmonizes exploration and constraint enforcement. Collectively, these designs enable robust and balanced RL finetuning for few-step T2MV diffusion models, yielding substantial gains in both per-view fidelity and cross-view consistency. Code is available at https://github.com/ZiyiZhang27/MVC-ZigAL.

cs.LG↗
arXiv2026

MosaicMem: Hybrid Spatial Memory for Controllable Video World Models

Video diffusion models are moving beyond short, plausible clips toward world simulators that must remain consistent under camera motion, revisits, and intervention. Yet spatial memory remains a key bottleneck: explicit 3D structures can improve reprojection-based consistency but struggle to depict moving objects, while implicit memory often produces inaccurate camera motion even with correct poses. We propose Mosaic Memory (MosaicMem), a hybrid spatial memory that lifts patches into 3D for reliable localization and targeted retrieval, while exploiting the model's native conditioning to preserve prompt-following generation. MosaicMem composes spatially aligned patches in the queried view via a patch-and-compose interface, preserving what should persist while allowing the model to inpaint what should evolve. With PRoPE camera conditioning and two new memory alignment methods, experiments show improved pose adherence compared to implicit memory and stronger dynamic modeling than explicit baselines. MosaicMem further enables minute-level navigation, memory-based scene editing, and autoregressive rollout.

cs.CV↗
arXiv2026

Rationale-Grounded In-Context Learning for Time Series Reasoning with Multimodal Large Language Models

The underperformance of existing multimodal large language models for time series reasoning lies in the absence of rationale priors that connect temporal observations to their downstream outcomes, which leads models to rely on superficial pattern matching rather than principled reasoning. We therefore propose the rationale-grounded in-context learning for time series reasoning, where rationales work as guiding reasoning units rather than post-hoc explanations, and develop the RationaleTS method. Specifically, we firstly induce label-conditioned rationales, composed of reasoning paths from observable evidence to the potential outcomes. Then, we design the hybrid retrieval by balancing temporal patterns and semantic contexts to retrieve correlated rationale priors for the final in-context inference on new samples. We conduct extensive experiments to demonstrate the effectiveness and efficiency of our proposed RationaleTS on three-domain time series reasoning tasks. We will release our code for reproduction.

cs.AI↗
arXiv2025

TwinSegNet: A Digital Twin-Enabled Federated Learning Framework for Brain Tumor Analysis

Brain tumor segmentation is critical in diagnosis and treatment planning for the disease. Yet, current deep learning methods rely on centralized data collection, which raises privacy concerns and limits generalization across diverse institutions. In this paper, we propose TwinSegNet, which is a privacy-preserving federated learning framework that integrates a hybrid ViT-UNet model with personalized digital twins for accurate and real-time brain tumor segmentation. Our architecture combines convolutional encoders with Vision Transformer bottlenecks to capture local and global context. Each institution fine-tunes the global model of private data to form its digital twin. Evaluated on nine heterogeneous MRI datasets, including BraTS 2019-2021 and custom tumor collections, TwinSegNet achieves high Dice scores (up to 0.90%) and sensitivity/specificity exceeding 90%, demonstrating robustness across non-independent and identically distributed (IID) client distributions. Comparative results against centralized models such as TumorVisNet highlight TwinSegNet's effectiveness in preserving privacy without sacrificing performance. Our approach enables scalable, personalized segmentation for multi-institutional clinical settings while adhering to strict data confidentiality requirements.

cs.CV↗
arXiv2025

WaterWave: Bridging Underwater Image Enhancement into Video Streams via Wavelet-based Temporal Consistency Field

Underwater video pairs are fairly difficult to obtain due to the complex underwater imaging. In this case, most existing video underwater enhancement methods are performed by directly applying the single-image enhancement model frame by frame, but a natural issue is lacking temporal consistency. To relieve the problem, we rethink the temporal manifold inherent in natural videos and observe a temporal consistency prior in dynamic scenes from the local temporal frequency perspective. Building upon the specific prior and no paired-data condition, we propose an implicit representation manner for enhanced video signals, which is conducted in the wavelet-based temporal consistency field, WaterWave. Specifically, under the constraints of the prior, we progressively filter and attenuate the inconsistent components while preserving motion details and scenes, achieving a natural-flowing video. Furthermore, to represent temporal frequency bands more accurately, an underwater flow correction module is designed to rectify estimated flows considering the transmission in underwater scenes. Extensive experiments demonstrate that WaterWave significantly enhances the quality of videos generated using single-image underwater enhancements. Additionally, our method demonstrates high potential in downstream underwater tracking tasks, such as UOSTrack and MAT, outperforming the original video by a large margin, i.e., 19.7% and 9.7% on precise respectively.

cs.CV↗
arXiv2025

RIS-Assisted Joint Sensing and Communications via Fractionally Constrained Fractional Programming

This paper studies an uplink dual-functional sensing and communication system aided by a reconfigurable intelligent surface (RIS), whose reflection pattern is configured to trade-off sensing and communication functionalities. Specifically, the Bayesian Cramér-Rao lower bound (BCRLB) for estimating the azimuth angle of a sensing user is minimized while ensuring the signal-to-interference-plus-noise ratio constraints for communication users. We show that this problem can be formulated as a novel fractionally constrained fractional programming (FCFP) problem. To deal with this nontrivial optimization problem, we extend a quadratic transform technique, originally proposed to handle optimization problems containing fractional structures only in objectives, to the scenario where the constraints also include ratios. First, we consider the case where the fading coefficient is known. Using the quadratic transform, the FCFP problem can be turned into a sequence of subproblems that are convex except for the constant-modulus constraints which can be tackled using a penalty-based approach. To further reduce the computational complexity, we leverage the constant-modulus conditions and propose a novel linear transform. This new transform enables the FCFP problem to be turned into a sequence of linear programming (LP) subproblems, which can be solved efficiently. Then, we consider the case where the fading coefficient is unknown. A modified BCRLB is used to make the problem more tractable, and the proposed quadratic transform based algorithm is used to solve the problem. Numerical results unveil nontrivial and effective reflection patterns that can be synthesized by the RIS to facilitate both communication and sensing functionalities.

eess.SP↗
arXiv2025

Capacity Bounds for Broadcast Channels with Bidirectional Conferencing Decoders

The two-user broadcast channel (BC) with receivers connected by bidirectional cooperation links of finite capacities, known as conferencing decoders, is considered. A novel capacity region outer bound is established based on multiple applications of the Csiszár-Körner identity. Achievable rate regions are derived by using Marton's coding as the transmission scheme, together with different combinations of decode-and-forward and quantize-bin-and-forward strategies at the receivers. It is shown that the outer bound coincides with the achievable rate region for a new class of semi-deterministic BCs with degraded message sets; for this class of channels, one-round cooperation is sufficient to achieve the capacity. Capacity result is also derived for a class of more capable semi-deterministic BCs with both common and private messages and one-sided conferencing. For the Gaussian BC with conferencing decoders, if the noises at the decoders are perfectly correlated (i.e., correlation is either 1 or -1), the new outer bound yields exact capacity region for two cases: i) BC with degraded message sets; ii) BC with one-sided conferencing from the weaker receiver to the stronger receiver. When the noises have arbitrary correlation, the outer bound is shown to be within half a bit from the capacity region for these same two cases. Finally, for the general Gaussian BC, a one-sided cooperation scheme based on decode-and-forward from the stronger receiver to the weaker receiver is shown to achieve the capacity region to within $\frac{1}{2}\log (\frac{2}{1-|λ|})$ bits, where $λ$ is the noise correlation. An interesting implication of these results is that for a Gaussian BC with perfectly negatively correlated noises and conferencing decoders with finite cooperation link capacities, it is possible to achieve a strictly positive rate using only an infinitesimal amount of transmit power.

cs.IT↗