Searcharxiv⌕ Search

arXiv subjects

Jie Tang

Publications and source records attributed to Jie Tang.

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

Reconfiguring Sparse Apertures: Partial-Mobility Planar FAS for 2-D DOA Estimation

Classical sparse arrays enlarge the sensing aperture under limited element and radio-frequency (RF) chain budgets, but their fixed geometries impose a persistent tradeoff between wide-sector identifiability and local angular resolution. This paper converts this static design into a two-state reconfigurable sparse aperture by allowing only part of a planar fluid antenna system (FAS) to move after an initial observation. A compact, Nyquist-spaced seed first provides an ambiguity-controlled acquisition geometry. The remaining fluid ports then move or switch to information-efficient refinement positions, and the measurements collected before and after reconfiguration are jointly processed. We formulate the receiver under fixed total numbers of ports and RF chains, a movement budget, and a common total-snapshot budget. A policy-level Fisher information matrix (FIM) identity accounts for the observation-dependent refinement geometry. For a single source, a relaxed planar D-optimal analysis gives the corner-favoring aperture law, while a finite-port certificate bounds the information retained under spacing, reachability, and partial-actuation constraints. For multiple sources, an aperture-and-conditioning surrogate generates feasible sparse layouts that are ranked and locally refined using the exact frame FIM with coarray, spacing, and movement regularization. Seed multiple signal classification (MUSIC) identifies a reliable angular basin, followed by joint concentrated-likelihood refinement over both states. Equal-budget simulations show that reconfiguring a sparse aperture converts available area and movement into lower angular error and retains much of the all-movable gain with fewer actuated ports. They also identify the operating boundary: if a full sparse aperture can remain permanently deployed, avoiding movement and snapshot splitting can be preferable.

eess.SP↗
arXiv2026

Sparse Channel Acquisition in Aging Fluid Antenna Systems: A Reliability-Calibrated Posterior Reconstruction Approach

Fluid antenna systems (FASs) exploit port-domain diversity within a compact aperture, but their gain can be lost if the receiver has to scan too many ports before data transmission. This paper studies sparse channel acquisition for aging FAS receivers, where the available information has mixed reliability: current observations are fresh but sparse, whereas historical channel state information (CSI) is dense but stale due to Doppler-induced temporal decorrelation. We formulate the problem from a communication perspective by linking aperture-field reconstruction to port-selection rate, outage, probing budget, and online inference latency. We then cast the acquisition task as reliability-calibrated posterior reconstruction, in which the sparse current observations serve as hard measurement evidence and historical CSI is reused only as soft temporal side information. The resulting reliability-controlled conditional diffusion posterior (DCP) framework conditions the denoising process on the sparse current observations, the observation mask, and the historical CSI, and refines the reverse trajectory through observation consistency, mismatch-aware temporal gating, spatial trust weighting, progressive temporal activation, and warm initialization. Analysis shows how aging and reconstruction error induce rate distortion and rate regret. Simulations across probing budgets, signal-to-noise ratios (SNRs), temporal correlations, observation patterns, ablation settings, and latency tests show consistent reconstruction gains with millisecond-level online inference latency, indicating that stale CSI can be beneficial when its strength, timing, and spatial support are explicitly controlled.

eess.SP↗
arXiv2026

Conformalized Large Language Models under Configuration Shift

Conformal prediction (CP) is a distribution-free framework for uncertainty quantification that has recently been adapted to large language models (LLMs), providing prediction sets with finite-sample coverage guarantees under exchangeability. Yet for LLMs, nonconformity scores are often induced by an inference pipeline, not just a fixed model, making them depend not only on the data distribution but also on configurable factors such as the prompt template, decoding parameters, and deployment setting. Since such configurations are routinely modified in practice but rarely treated as a source of shift, their impact on CP validity remains poorly understood. We call this \emph{configuration shift} and study it systematically along three axes: prompt template, decoding temperature, and weight quantization. In a broad empirical study spanning $9$ LLMs, $4$ datasets, and $4$ nonconformity scores, we find that configuration shift consistently erodes CP validity, often driving empirical coverage below the target. By contrast, efficiency is largely preserved: valid prediction sets remain close in size to the i.i.d. baseline. We derive coverage lower bounds that attribute this loss to a discrepancy between calibration and test score distributions, and use their finite-sample plug-in versions as empirical diagnostics of shift severity. We further show that these findings lead to practical mitigations: bound-inspired recalibration is effective with limited test examples, while fragility-aware calibration ensembling recovers much of the lost coverage without test data.

cs.LG↗
arXiv2026

AWM: Answerable Working Memory for Long-Document VQA Agents

Long-document visual question answering increasingly relies on VLM agents that retrieve candidate pages, inspect page images, write findings to working memory, and synthesize answers. Working memory should carry answer-supporting evidence across page inspections for later grounded answering, yet existing evaluation mainly checks final-answer correctness and evidence-page access. This creates a memory-quality blind spot: an agent may reach the right page and answer correctly while leaving behind memory too generic or incomplete to support answering once page context is removed. We introduce \emph{memory-only answerability}, a diagnostic that asks whether a reader can answer from the question and terminal working memory alone. Building on this diagnostic, \emph{Answerable Working Memory} (AWM) treats terminal working memory as an answerable evidence artifact, and AWM-GRPO incorporates this signal into the GRPO reward while preserving final-answer priority. Under GRPO, this reward assigns higher advantages to answer-correct trajectories whose terminal working memory remains answerable. On \textsc{MMLongBench-Doc}, even when gold evidence pages are provided, 42.5\% of correct answers still cannot be answered from terminal working memory alone. AWM-GRPO improves final-answer accuracy over the RAG baseline by 8.1 and 11.9 points on \textsc{MMLongBench-Doc} and \textsc{LongDocURL} and reduces the memory-missing-correct rate by 2.7 points over answer-only GRPO.

cs.CL↗
arXiv2026

Electromagnetic Twin: From Sparse Measurements to Persistent Wireless Intelligence

Radio maps and channel knowledge maps provide reusable propagation knowledge, but a stored map can become locally outdated after persistent changes to doors, partitions, furniture, large equipment, or infrastructure. Motivated by digital-twin state synchronization, we develop an electromagnetic twin that recurrently converts sparse channel measurements into a persistent wireless state, exposes that state to communication queries, and uses its uncertainty to request subsequent measurements. The twin tracks persistent or semi-persistent propagation changes rather than transient human motion or fast fading. Each update is a constrained inverse problem that combines the previous map, a scene graph, and an imperfect physics prior. Measurement consistency with a confidence-calibrated radius and physical gain bounds enforce feasibility, while scene-aware spatial regularization and selective temporal memory preserve propagation boundaries and unchanged regions. Successive convex approximation (SCA), majorization--minimization alternating direction method of multipliers (MM-ADMM), and a low-complexity primal--dual hybrid gradient (LC-PDHG) mode realize the framework at different computational scales, and a local perturbation analysis gives an explicit multi-update tracking recursion. The updated state supports access-point association and codebook beam selection, while change-weighted A-optimal design closes the measurement--update--query loop. Over 500 realizations, the recurrent update remains stable through eight persistent scene events and raises best-beam accuracy from $76.64\%$ to $90.83\%$ as measured-location density grows from $2\%$ to $16\%$.

eess.SP↗
arXiv2026

Electromagnetic Twin: Completing the Wireless World from Sparse Channel Evidence

Acquiring dense channel information over many locations and beams incurs considerable pilot and processing overhead. Radio maps and channel knowledge maps (CKMs) reduce this overhead by reusing site-specific channel information, but their contents must be refreshed when new measurements or environmental observations become available. This paper introduces an \emph{electromagnetic twin} as an updatable digital representation that uses sparse channel evidence to reconstruct the wireless state requested by communication queries. Rather than replacing radio maps or CKMs, the twin uses a CKM as channel memory, combines it with registered scene information, and regenerates its outputs after each evidence update. We instantiate this idea by completing a two-dimensional channel-gain field from sparse samples and an incomplete floor plan. A learned RF completion backbone recovers the main propagation structure, and a lightweight residual adapter tests whether frozen CLIP features provide useful side information. With $4\%$ measured locations and $55\%$ missing semantic objects, the RF backbone attains $4.44$ dB RMSE, compared with $8.29$ dB for CKM interpolation and $8.38$ dB for an incomplete physics prior. Residual adaptation reduces RMSE by a paired mean of $0.135$ dB (95\% confidence interval: $0.100$--$0.169$ dB), but a same-capacity random-feature control is statistically indistinguishable from the CLIP-conditioned adapter. The results therefore support the measurement--update--query loop and lightweight residual correction, while avoiding an unsupported attribution of the correction to visual semantics.

eess.SP↗
arXiv2026

From Sparse Probes to Sum-Rate Maximization: Electromagnetic Twin Beamforming

Probing every candidate location at each wireless-map update is costly. This letter develops electromagnetic-twin (ET) beamforming that converts sparse spatial probes into a sum-rate decision. The ET stores a location-dependent angular power spectrum, updates its innovation through graph-regularized estimation, and queries user covariances for projected beam optimization. A rate-sensitivity bound selects subsequent probes by query-weighted posterior-variance reduction. With 1\% random probes, ET beamforming achieves $2.69$ bit/s/Hz versus $1.12$ for a static channel knowledge map; with 7\% query-aware probes, it reaches $3.03$ bit/s/Hz, within 3.4\% of perfect-covariance beamforming.

eess.SP↗
arXiv2026

TideRL: Boosting Agentic RL Goodput with Readiness-Aware Scheduling

Reinforcement learning (RL) for large language models is moving toward multi-turn agentic workloads, where rollout tasks repeatedly pause for external environments, resume with growing contexts, and finish at highly variable times. In this setting, RL training goodput, measured by training throughput, matters more than raw GPU occupancy: GPU waiting and repeated prefill recomputation are pure overhead. We present TideRL, a readiness-aware elastic RL system with Continuous Task Batching, Resource-Aware Ref-Actor Pipelining, and Elastic Resource Scaling. CTB preserves useful rollout state, $\textrm{RA}^2\textrm{P}$ selects between decoupled streaming and colocated aggregation from the ready backlog and arrival interval, and ERS moves ranks between rollout and training using the same readiness signals. Across text-only and multi-modal agentic workloads, TideRL improves RL training goodput by up to 5.6$\times$ over synchronous baselines and over 33% over asynchronous baselines, while reaching similar task performance. It also improves KV cache hit rate by 1.58$\times$, reduces per-step training time by up to 44.3%, and cuts total waiting time by up to 77.6%.

cs.LG↗
arXiv2026

Timescale-dependent Optical Variability of Turn-on Changing-look Active Galactic Nuclei

Changing-look active galactic nuclei (CL AGNs) provide a valuable opportunity to study optical variability associated with changes in accretion state. We investigate the optical variability of turn-on CL AGNs, selected through the emergence or strengthening of broad emission lines, using g-band light curves from the Zwicky Transient Facility. For a final sample of 106 objects, we measure structure-function amplitudes at three fixed rest-frame timescales, SF30, SF150 and SF300, and examine their dependence on optical luminosity, Eddington ratio, black-hole mass and rest-frame wavelength using Spearman-rank correlations and multivariate regressions. The strongest trends are found on monthly timescales. SF30 is significantly anti-correlated with both optical luminosity and Eddington ratio, with Spearman coefficients of rho = -0.39 and rho = -0.33, respectively. These negative trends remain in multivariate regressions after accounting for black-hole mass and rest-frame wavelength. The dependence weakens at longer timescales: SF150 shows only weak evidence for luminosity or Eddington-ratio dependence, while SF300 shows no robust dependence. Black-hole mass shows no robust single-parameter correlation, and its multivariate coefficients depend on the model form, indicating that its apparent role is affected by covariance with luminosity and Eddington ratio. The rest-frame wavelength term is generally negative, but cannot be separated from redshift in this single-band analysis. These results suggest that monthly optical variability in turn-on CL AGNs is more closely linked to bright-state accretion properties than variability on half-year to year-long timescales.

astro-ph.GA↗
arXiv2026

SurveyReview: A Reviewer-Aligned Benchmark for Survey Evaluators

The rapid advancement of large language models has transformed survey writing from a months-long manual effort into an automated process. As generation scales, reliable evaluation becomes the bottleneck, and LLMs are increasingly used as survey evaluators. However, existing approaches largely rely on off-the-shelf LLM-as-a-judge methods without systematic alignment to human reviewers, and there remains a lack of systematic frameworks for quantifying alignment with human reviewers. To address this gap, we propose SurveyReview, a reviewer-aligned, multi-dimensional benchmark and dataset for survey evaluation. We collect and annotate 675 survey papers with 1,630 review reports. We structure authentic peer-review reports by converting free-form comments into four-dimensional scores (Readability, Criticalness, Comprehensiveness, Structure) paired with supporting rationales. We further release standardized train/test splits and an evaluation protocol to measure alignment between automatic evaluators and human reviewers. To validate the benchmark, we develop SurveyAlign, a strong baseline evaluator by fine-tuning Qwen3-32B with LoRA on our annotated data, augmented with external knowledge for knowledge-intensive dimensions. On the test set, SurveyAlign substantially improves reviewer alignment over prompt-based judging with GPT-5.2, reducing average MSE from 2.28 to 1.38 and MAE from 1.15 to 0.69 across all four dimensions. Our contributions are twofold: (1) we establish the first multi-dimensional, reviewer-aligned dataset with a reproducible evaluation framework for survey reviewing; (2) we develop a strong baseline evaluator that substantially improves alignment with human reviewers, providing a competitive reference for future research. Our code and data are available at https://surveyreview.github.io

cs.CL↗
arXiv2026

SCAIL-2: Unifying Controlled Character Animation with End-to-End In-Context Conditioning

Controlled character animation aims to transfer motion from a driving sequence to a reference character. Prior works heavily rely on intermediate representations, such as pose skeletons for motion and masked backgrounds for environment, inevitably resulting in information loss. In this work, we present SCAIL-2, a framework that adopts an end-to-end driving paradigm by directly concatenating latent visual information to the model's input sequence. We enable end-to-end training through a data synthesis pipeline that produces MotionPair-60K, a curated dataset for several character animation subtasks. We unify the subtasks using decoupled conditions to accommodate different driving patterns, facilitated by In-Context Mask Conditioning and Mode-Specific RoPE, which provide soft guidance beyond textual instructions and visual information. To address synthetic discrepancy in detailed regions, we propose Bias-Aware DPO to construct preference items to mitigate the errors. Extensive experiments demonstrate that our method achieves state-of-the-art performance across various character animation tasks. Code, model weights, and a large subset of the dataset are available at: https://teal024.github.io/SCAIL-2/.

cs.CV↗
arXiv2026

Enabling Proactive Spoken Turns via a Generalized Style-Aware Full-Duplex Framework

Compared with half-duplex dialogue systems where the system waits for user turn completion before it responds, natural full-duplex dialogue systems require agents to act proactively in real time, including timely interruptions and backchannels. This creates a key challenge: improving turn timing without sacrificing response quality. To address limitations in realistic proactive turn-taking, we build a generalized style-aware full-duplex framework with three key components. Firstly, we propose LPS-TC, a Lightweight Proactive Speech Turn Controller for plug-and-play integration. It features a fine-grained action space covering both reactive and proactive turn behaviors, enabling half-duplex models with full-duplex capabilities and enhancing existing full-duplex models with superior timing control. Secondly, we construct WildTurn, a large-scale, real-world English dataset containing approximately 2,981 hours of filtered multi-turn stereo conversations from face-to-face and telephone conversations, annotated with five turn-taking and five backchanneling styles. Trained on WildTurn, LPS-TC exhibits rich spoken dynamics that are not captured by existing static full-duplex benchmarks. Thirdly, we introduce a two-tier evaluation scheme that assesses both chunk-level timing precision and turn-level interaction quality under realistic streaming constraints. Our experiments, integrating LPS-TC with half-duplex models like Qwen2.5-Omni and full-duplex models like Freeze-Omni, showcase its superior performance in timing appropriateness and response quality. Our framework also demonstrates fine-grained style controllability and strong generalizability, enabling more natural and human-like spoken interactions.

cs.CL↗
arXiv2026

Memory for Large Language Models

Memory has evolved into a foundational architectural dimension in large language models (LLMs), shifting from an implicit byproduct of computation to a spectrum of explicit, controllable mechanisms. While recent advances introduce diverse strategies---spanning transient attention, recurrent state dynamics, parameter-efficient adaptations, and scalable lookup storage---this rapid evolution has led to a highly fragmented research landscape. In this survey, we present a systematic, architecture-centric taxonomy of memory in LLMs. Our framework characterizes memory along three orthogonal axes: representation (implicit versus explicit), update dynamics (offline versus online), and persistence (short-term versus long-term). We further formalize the granular mechanisms dictating memory writing, routing, state transitions, and consolidation. This unified perspective elucidates the conceptual boundaries between computation-coupled and independently addressable memory, effectively bridging disparate architectural paradigms. Additionally, we critically analyze hybrid memory architectures, system-level efficiency trade-offs, and multi-dimensional evaluation methodologies. By consolidating these scattered advancements into a cohesive framework, this survey charts the trajectory of memory-centric LLM design and provides a principled foundation for future innovations in scalable and adaptive language modeling.

cs.CL↗
arXiv2026

Delving into Latent Spectral Biasing of Video VAEs for Superior Diffusability

Latent diffusion models pair VAEs with diffusion backbones, and the structure of VAE latents strongly influences the difficulty of diffusion training. However, existing video VAEs typically focus on reconstruction fidelity, overlooking latent structure. We present a statistical analysis of video VAE latent spaces and identify two spectral properties essential for diffusion training: a spatio-temporal frequency spectrum biased toward low frequencies, and a channel-wise eigenspectrum dominated by a few modes. To induce these properties, we propose two lightweight, backbone-agnostic regularizers: Local Correlation Regularization and Latent Masked Reconstruction. Experiments show that our Spectral-Structured VAE (SSVAE) achieves a $3\times$ speedup in text-to-video generation convergence and a 10\% gain in video reward, outperforming strong open-source VAEs. The code is available at https://github.com/zai-org/SSVAE.

cs.CV↗
arXiv2026

Vision2Web: A Hierarchical Benchmark for Visual Website Development with Agent Verification

Recent advances in large language models have improved the capabilities of coding agents, yet systematic evaluation of complex, end-to-end website development remains limited. To address this gap, we introduce Vision2Web, a hierarchical benchmark for visual website development, spanning from static UI-to-code generation, interactive multi-page frontend reproduction, to long-horizon full-stack website development. The benchmark is constructed from real-world websites and comprises a total of 193 tasks across 16 categories, with 918 prototype images and 1,255 test cases. To support flexible, thorough and reliable evaluation, we propose workflow-based agent verification paradigm based on two complementary components: a GUI agent verifier and a VLM-based judge. We evaluate multiple visual language models instantiated under different coding-agent frameworks, revealing substantial performance gaps at all task levels, with state-of-the-art models still struggling on full-stack development.

cs.SE↗
arXiv2026

SCALECUA: Scaling Computer Use Agents with Verifiable Task Synthesis and Efficient Online RL

Computer use agents (CUAs) are emerging as a powerful interface for automating complex digital workflows through visual perception and GUI execution. Online reinforcement learning with verifiable rewards (RLVR) has emerged as a key direction for scaling their capabilities. However, this paradigm is bottlenecked by verifiable data scarcity and online RL inefficiency. To break these barriers, we introduce ScaleCUA, a unified framework that scales online RL for CUAs via verifiable task synthesis and efficient training. At the data level, we design VeriGen, an end-to-end framework for generating verifiable RL tasks through iterative docker interactions and a multi-agent feedback loop. Scaled to 100+ concurrent agent workers via a shared docker interaction probe, this pipeline produces 24K+ verifiable tasks and nearly 3K high-quality RL tasks. To maximize sample efficiency, we propose Frontier Sampling, which tracks per-task capability and allocates rollouts to the current learning frontier. On the training side, we further design Visual Context Segmentation, a sliding window over recent visual context that balances rollout and training-engine pressure, yielding a 2.83x training speedup over step-wise decomposition. Together, ScaleCUA achieves 68.7% on OSWorld and 54.0% on ScienceBoard, establishing new state-of-the-art performance among open-source computer use agents. Code, models, and datasets are available at https://github.com/THUDM/SCALE-CUA.

cs.AI↗
arXiv2026

Rotatable Antenna-Empowered Wireless Networks: A Tutorial

Non-fixed flexible antenna architectures, such as fluid antenna system (FAS), movable antenna (MA), and pinching antenna, have garnered significant interest in recent years. Among them, rotatable antenna (RA) has emerged as a promising technology for enhancing wireless communication and sensing performance through flexible antenna orientation/boresight rotation. By enabling mechanical or electronic boresight adjustment without altering physical antenna positions, RA introduces additional spatial degrees of freedom (DoFs) beyond conventional beamforming. In this paper, we provide a comprehensive tutorial on the fundamentals, architectures, and applications of RA-empowered wireless networks. Specifically, we begin by reviewing the historical evolution of RA-related technologies and clarifying the distinctive role of RA among flexible antenna architectures. Then, we establish a unified mathematical framework for RA-enabled systems, including general antenna/array rotation models, as well as channel models that cover near- and far-field propagation characteristics, wideband frequency selectivity, and polarization effects. Building upon this foundation, we investigate antenna/array rotation optimization in representative communication and sensing scenarios. Furthermore, we examine RA channel estimation/acquisition strategies encompassing orientation scheduling mechanisms and signal processing methods that exploit multi-view channel observations. Beyond theoretical modeling and algorithmic design, we discuss practical RA configurations and deployment strategies. We also present recent RA prototypes and experimental results that validate the practical performance gains enabled by antenna rotation. Finally, we highlight promising extensions of RA to emerging wireless paradigms and outline open challenges to inspire future research.

cs.IT↗
arXiv2026

Single-Rollout Asynchronous Optimization for Agentic Reinforcement Learning

Reinforcement learning (RL) is becoming increasingly important for post-training large language models (LLMs). Previous RL pipelines for LLMs were mostly synchronous and batch-interleaved, which is inefficient for long-horizon agentic tasks. Recently, asynchronous RL has emerged as a more efficient alternative by updating the model as rollouts arrive. However, existing asynchronous RL systems often emphasize throughput, while leaving training stability and task effectiveness largely underexplored. For example, a key challenge is that group-wise sampling in the widely-used GRPO framework does not naturally fit asynchronous agentic training. In this paper, we present Single-rollout Asynchronous Optimization (SAO) to address the stability and off-policy challenges in asynchronous RL. To reduce off-policy effects and improve generalization, we replace group-wise sampling with single-rollout sampling, that is, using one rollout per prompt. We further improve this single-rollout strategy with practical value-model training designs. To improve optimization stability, we introduce a strict double-side token-level clipping strategy. SAO is able to train stably for one thousand steps and consistently outperform GRPO and its variants on agentic coding and reasoning benchmarks, such as SWE-Bench Verified, BeyondAIME, and IMOAnswerBench. We also demonstrate that single-rollout RL is particularly effective in a simulated online learning setting, where the model must adapt to changing evolving environments. To this end, SAO is successfully deployed in the agentic RL pipeline for training the open GLM-5.2 model (750B-A40B).

cs.LG↗