arXiv subjects

Yuchen Zhang

Publications and source records attributed to Yuchen Zhang.

At least 19 recordsLinked to original sources

arXiv2026

UniTele: Learning Device-Conditioned Degradations for Unified Smartphone Telephoto Super-Resolution

Super-telephoto imaging on mobile phones suffers from complex, device-dependent degradations that are poorly captured by conventional handcrafted pipelines. Collecting large-scale aligned training pairs across different phones and diverse scenes, however, is both costly and difficult to scale. We therefore study whether device-specific degradations can be learned from only a small number of physical captures and transferred for training a unified super-resolution (SR) model. Our key idea is to factor degradation learning into two learnable components: one captures the device-specific imaging characteristics under well-resolved conditions, while the other captures the additional information loss and artifacts that arise in super-telephoto imaging. We build a paired benchmark across three phone brands (Huawei, Apple, and vivo) by photographing the same content at two display sizes and zoom settings. The larger displayed image is captured at the phone's maximum optical zoom setting, while the smaller image is photographed directly at the phone's highest native zoom setting through the phone's native imaging pipeline. The benchmark contains 519 training and 60 held-out pairs. We use the benchmark to learn device-specific degradation, then apply it to external images to synthesize thousands of additional pairs to train our UniTele SR model. Relative to the pretrained Qwen and FLUX backbones before adaptation, the full UniTele training recipe substantially improves paired restoration fidelity and perceptual similarity on the held-out benchmark and yields markedly more stable and visually faithful restorations.

cs.CV↗

arXiv2026

ARISE: Authentic Reference-Informed Stability Enhancement for Continual Multimodal Forgery Detection

Multimodal forgery detection is commonly studied under a static assumption: training attacks are expected to approximate those encountered after deployment. This assumption becomes increasingly fragile as generators, editing systems, and multimodal large language models continually introduce new forgery mechanisms and forensic evidence. We formulate this problem as attack-incremental continual multimodal forgery detection, where the binary real/fake interface remains fixed while emerging attacks must be learned without forgetting previous ones. We introduce Fable-Bench, which organizes attack evolution through complementary joint-modality, text-side, and image-side protocols. We further propose ARISE (Authentic Reference-Informed Stability Enhancement), motivated by the open-ended forged class and the comparatively reusable geometry of authentic content. ARISE fits a real-only Gaussian mixture model to projected authentic replay representations as a compact historical reference. Meanwhile, the proposed Gaussian-Responsibility-Aware Stability Preservation (GRASP) derives a sample-adaptive anisotropic metric that strongly constrains historically stable representation directions while retaining plasticity elsewhere. Our Vulnerability-Aware Replay (VAR) further estimates stage-wise vulnerability from authentic-density degradation, forged-to-authentic boundary confusion, and authentic-mode occupancy drift, then assigns bounded replay emphasis to fragile stages. All adaptive signals are obtained from training-time replay statistics, while inference uses only the original multimodal backbone and binary classification head. Experiments across three attack-evolution protocols and multiple detector architectures demonstrate improved historical retention and final average performance over representative continual-learning baselines.

cs.CV↗

arXiv2026

Preserve to Expand: Source-Aware Adaptation and Model Merging for Multilingual LLM-Based ASR

Expanding multilingual ASR systems to new languages is challenging when adaptation must preserve previously supported languages without replaying source-language training data. We study this problem in connector-based Speech LLMs, where a trainable connector maps speech representations into a frozen large language model. We evaluate target-update construction and merging strategies for adapting the connector while preserving source-language performance. In single-target expansion, target-only fine-tuning and standard TIES merging learn new languages but cause severe source-language forgetting, while SSU-Wanda+TIES gives the best source-preserving trade-off. In multi-target expansion, update merging provides stable source retention, whereas recursive expansion can improve target adaptation under favorable orders but is highly order-sensitive. These results show that multilingual ASR expansion requires balancing target acquisition, source retention, and interactions among target-language updates.

cs.CL↗

arXiv2026

EvoSteer: Online Self-Evolving Graph Orchestration via Reference-Anchored Credit Assignment

In recent years, LLM-based multi-agent systems have been widely applied to orchestrate tool-using agents into executable communication graphs. However, existing self-evolving orchestration still faces key challenges, including post-hoc evolution that revises the team only after the trajectory ends, credit diffusion that gives every action the same terminal advantage under confounded baselines, and skill admission that is uncalibrated and never retired. To address these challenges, we propose EvoSteer, a new paradigm of Online Self-Evolving Graph Orchestration -- the orchestrator builds a running team and repairs its plausible but failing steps from execution features and a learned value estimate. To support this paradigm, we introduce Anchored Trajectory Balance (AnchorTB), a regression-style flow-matching loss that assigns each orchestration action a coefficient by balancing subtrajectories against a frozen reference. Built on the learned flow, we further propose Validated Skill Admission, in which a candidate skill is tried before promotion and promoted only if paired evidence passes a sequential test under a shared nominal testing budget. Moreover, AnchorTB combines measured task-level reference reward statistics with prefix-dependent corrections. Experimental results on twelve datasets show that EvoSteer significantly outperforms baselines across question answering, mathematical reasoning, code generation, and interactive decision making. Our code is available at https://github.com/beita6969/evosteer.

cs.AI↗

arXiv2026

Octonionic and Monogenic Completions of Harmonic Functions

Let $Ω\subset\mathbb{R}^n$ be a domain, with $n\ge2$. We characterize the real-valued harmonic functions that can occur as prescribed real components of monogenic functions. For any fixed nonzero real Clifford module $M$ and any fixed unit vector $s\in M$, a smooth real-valued function $h$ can be written as $h=\langle F,s\rangle$ with $F$ monogenic if and only if the $(n-1)$-form $*dh$ is exact. Consequently, every real-valued harmonic function on $Ω$ admits such a representation if and only if $H_{\mathrm{dR}}^{n-1}(Ω)=0$. In dimension eight, the same exactness criterion characterizes the real parts of octonionic regular functions and yields the corresponding equivalence with $H_{\mathrm{dR}}^7(Ω)=0$. The proofs use global solvability of the Poisson equation and elementary identities for differential forms.

math.CV↗

arXiv2026

Handover-Aware Power Minimization for Networked LEO Satellite Communications: Joint Cooperative Beamforming and Scheduling

Networked low Earth orbit (LEO) satellite constellations enabled by inter-satellite links offer a promising path toward ubiquitous broadband non-terrestrial services, as cooperative transmission increases the received power and suppresses interference. However, fast orbital motion induces frequent handovers, which cost energy and transmission time, while the limited number of radio-frequency chains per satellite couples user scheduling with cooperative beamforming. In this paper, we investigate handover-aware power-efficient downlink transmission in networked LEO systems under statistical channel state information. Specifically, we first introduce a two-segment frame structure that separates the handover procedure from user-plane transmission, and propose a power consumption model that captures the energy to establish each new satellite-user link and the transmission time lost during its handover. We then formulate a per-frame network-wide power minimization problem that jointly optimizes cooperative beamforming and user scheduling, and develop an iterative algorithm that combines a reweighted $\ell_2$ surrogate with a penalty-based relaxation and a fractional-programming inner loop, which yields convex subproblems. Simulation results with Red Sea ship traffic demonstrate that, at $25$~Mbps per user, the proposed design needs 14\% less power than distance-based cooperative scheduling and 15\% less than a joint design that ignores the established links, while non-cooperative transmission meets the rate requirements in only 14\% of the frames.

eess.SP↗

arXiv2026

HandsOn: A Large-Scale Simulation Benchmark for Generalizable Dexterous Manipulation

General-purpose manipulation requires reliable behavior across embodiments, objects, and task demands. We introduce HandsOn, a composable simulation platform and benchmark for cross-embodiment dexterous manipulation. It supports five robotic arms, six dexterous hands, and approximately 200 rigid, articulated, and deformable assets in randomized scenes. The benchmark pairs 18,000 curated human demonstrations with 66 human-verified tasks: Core-50 supports demonstration-based learning and execution tests, while Transfer-16 evaluates asset and relation transfer, skill composition, and deformable manipulation without target-task training demonstrations. Qualitative real-robot rollouts illustrate sim-to-real transfer. Across diffusion, vision-language-action, world-action, and prompted VLM-based methods, the best trained multi-task policy achieves 29.1% Core-50 success. Without benchmark finetuning, GPT-6 using move_eef for palm-and-finger control achieves 5.2% on Core-50 and 1.9% on Transfer-16 over 660 trials. Trained-policy stage diagnostics expose failures in contact establishment, action coordination, and maintaining achieved goals, with additional sequencing errors in composed tasks. HandsOn provides a shared basis for investigating manipulation generalization and the limits of reliable physical interaction.

cs.RO↗

arXiv2026

CausalEvo: Causal Skill Self-Evolution for Health LLM Agents

Large language model (LLM) agents for health can accumulate reusable textual skills from task feedback, such as calculation criteria, measurement time points, and formula units. Deciding which skills to retain is consequential: useful rules improve later answers, while faulty generalizations propagate errors. We propose CausalEvo, a method for causal skill self-evolution that estimates the relative effects of deliberately changed skill sets while holding the model weights, feedback cases, and rule count fixed. At each evolution step, a frozen LLM proposes candidate rules, which are pooled with the incumbent skills so that the current skill set remains a feasible choice. Paired skill replacements against shared background rules are scored on the same feedback cases; least squares converts the score differences into relative skill values, and the top-ranked rules are retained, favoring incumbents on ties. A decision-loss bound separates candidate support, approximation, and estimation errors. On six MedCalc-Bench-Verified tasks with 120 held-out cases, CausalEvo achieves 60.00% accuracy, compared with 55.83% for SkillOpt and 51.67% without skills. Across the recorded training configurations, which use different optimization budgets, it consumes 55.6% fewer training tokens than SkillOpt. Gains concentrate in particular calculations alongside regressions on others, and paired accuracy intervals include zero. CausalEvo turns task feedback into controlled decisions about skill retention and replacement. Project repository: https://github.com/txbrook/CausalEvo

cs.LG↗

arXiv2026

Historical-Feature Residual Adaptation for Sequential Screening

Sequential screening policies must use acquired feedback effectively under a fixed experiment budget. We study acquisition-time residual adaptation: a small ranking correction built from fixed historical features and anchored to base scores recorded before feedback is revealed. The interface accepts a history-conditioned scorer and a candidate feature table; its primary instantiation augments AssayFormer's ten-dimensional representation with 64 historical directions while retaining the pretrained model. In retrospective evaluation on 20 official-test screens from 15 recorded sources, a comparison fixed before target-outcome access increased source-macro fraction of hits recovered from 22.58% to 28.61%, exceeding all six prespecified comparators. A same-feature ordinary ranking head achieved 25.73%. Supplementary strong logistic, rank-fusion, and unprojected controls achieved 26.42%, 26.02%, and 26.78%, respectively. A separate model-independent pilot on 13 reused development screens reconstructs the features without AssayFormer assets. Across three seeds, residual adaptation improves random-forest hit recovery from 28.87% to 31.27%, while logistic recovery changes from 30.22% to 29.78%. The results show effective pretrained-policy adaptation and useful transfer to an independent nonlinear base, with gains that depend on the base model.

cs.LG↗

arXiv2026

Beamforming Design for Stem-Connected Microwave Linear Analog Computer (MiLAC)-Aided Multiuser MISO Downlinks

A microwave linear analog computer (MiLAC) is a tunable microwave network that leverages wave propagation to perform computation directly in the analog domain. MiLAC finds promising applications in beamforming, where it can map the data streams to the antennas without a digital beamformer, using one radio-frequency chain per stream. A fully-connected MiLAC needs a number of tunable components that grows quadratically with the number of antennas, whereas a stem-connected MiLAC needs only a linearly growing number and still achieves the point-to-point capacity. For the multiuser multiple-input single-output downlink, we prove that the stem-connected MiLAC attains the maximum sum-rate of the fully-connected MiLAC whenever $N\ge 2K-1$, where $N$ and $K$ denote the numbers of antennas and users. The proof shows that restricting the design to beamformers with orthonormal columns, which the stem-connected MiLAC realizes in closed form, loses no sum-rate. Based on this restriction, we develop, for every $N\ge K$, a weighted minimum mean-square error algorithm with a Riemannian beamformer update and prove that its accumulation points are stationary. For the bounded, discrete susceptances of practical components, we then propose an alternating refinement that adjusts one susceptance at a time through rank-one updates and never decreases the sum-rate. With $64$ antennas, four users, and ideal components, the stem-connected MiLAC attains the fully-connected sum-rate with $4.5$ times fewer tunable components. With $3$-bit susceptances at a signal-to-noise ratio of $10$~dB, the refinement retains $90\%$ of this sum-rate.

eess.SP↗

arXiv2026

TartanMatch: Towards Universal Dense Correspondence Across Modalities

Robotic platforms combine multi-modal sensors to operate across diverse conditions, but most dense correspondence methods are limited to RGB-RGB or a small set of modality pairs. Training a model for each modality pair scales poorly with the number of sensors and forgoes potential positive transfer across modality pairs. We introduce TartanMatch, a unified model for dense correspondence across all 25 ordered pairs among RGB, event, thermal, depth, and projected LiDAR. TartanMatch adapts an RGB-pretrained correspondence model to all modality pairs within one shared network using lightweight modality adaptation and a mixture of real and synthetic data. On zero-shot real-world benchmarks, TartanMatch reduces Endpoint Error (EPE) over the strongest baselines by 61.9% for cross-modal matching and 49.6% for same-modal matching. We further show that joint training across all modality pairs improves average matching accuracy over pair-specific training, especially on challenging cross-modal pairs. These results demonstrate that a single model achieves broad multi-modal coverage and state-of-the-art matching accuracy.

cs.CV↗

arXiv2026

Topological Spatiotemporal Mode-Locked Lasers

Achieving robust high-dimensional solitons both in optics and Bose-Einstein condensates remains a great challenge, due to unpredictable high-order perturbations. Here we propose and demonstrate a novel topological spatiotemporal mode-locked fiber laser empowered by mode-division multiplexing, for precisely controlling of the spatiotemporal solitons. By introducing time delays between the individual transverse mode paths, a topological spatiotemporal lattice is formed in the multimode cavity, which in turn excites periodic spatiotemporal solitons. These solitons are stable against the perturbations, attributed to periodic potentials induced by the nonreciprocal dissipative coupling interactions. Moreover, under the thermodynamic limit, the solitons condense into nested-type spatiotemporal solitons, a phenomenon analogous to the Bose-Einstein condensate of matter waves. Our results indicate that topological spatiotemporal mode-locked lasers not only constitute a controllable source of robust spatiotemporal light bullets, but also provide a valuable platform for exploring nonlinear high-dimensional soliton dynamics and non-Hermitian topological physics.

physics.optics↗

arXiv2026

InterCorrect: Intersection-Aware Correction of Demographic Model Merging for Fair ASR

Automatic Speech Recognition (ASR) systems often show uneven performance across demographic groups, and errors can be especially difficult to address for speakers belonging to multiple demographic groups. This work studies demographic-aware model merging for fair Speech-LLM-based ASR. Starting from a SLAM-ASR-based model, we fine-tune only the connector on demographic-specific subsets and merge the resulting subgroup-adapted connectors into a global model. We then identify critical cross-axis demographic pairs using subgroup WER and task-vector conflict, and apply intersection-specific correction vectors to the global merged model. Experiments on Fair-Speech show that global demographic merging improves overall WER over the base model, while intersection correction provides additional gains for several merging strategies. In particular, TIES with WER-based correction achieves the best overall WER, reducing it from 7.38\% to 5.13\%. Subgroup and disparity analyses further show that the proposed approach improves performance across demographic axes, while highlighting that lower average WER does not always imply reduced subgroup disparity.

cs.CL↗

arXiv2026

CARD: Cross-component Audio Representation Distillation for Encoder-Free Audio Captioning

Modern automated audio captioning systems pair a frozen audio encoder with a large language model (LLM) via a trainable projector, incurring the encoder's inference cost and bottlenecking the model through its fixed acoustic features. We present CARD, an encoder-free audio captioning model that removes the encoder at inference: a 13.2M projector feeds a frozen LLM with merged LoRA adapters, while the teacher used to train it is discarded. CARD distills a pre-trained audio teacher (CLAP-HTSAT) into the model, but rather than injecting it into the LLM alone, it routes the teacher's representations across components: perceptual stages to the projector and semantic stages to the LLM. This placement improves CIDEr-D by +11.9 over an LLM-only distilled model on AudioCaps and by +5.0 on Clotho, reaching 55.4 against a 66.4 encoder-kept upper bound with no encoder at inference, showing that where a teacher's knowledge is placed matters as much as its presence.

cs.SD↗

arXiv2026

A Koszul complex in quaternionic analysis and its applications

Let $n\geqslant 1, Ω\subset\mathbb{H}^n $ be a domain. We construct a Koszul-type complex for the ideal sheaf $\mathcal{I}_X^{(k)}$ of $k$-regular functions vanishing on $X=\{(q_0, q_1, \cdots, q_{n-1})\in Ω: q_0=0\}$ in several quaternionic variables: $$0\to \mathcal{R}^{(k+2)}\xrightarrow{\widetilde{\mathscr{L}}^{(k)}} \mathcal{R}^{(k+1)}\oplus\mathcal{R}^{(k+1)}\xrightarrow{\mathscr{L}^{(k)}} \mathcal{I}_X^{(k)}\to 0,$$ where $k\geqslant 0$, $\mathcal{R}^{(k)}$ is the sheaf of $k$-regular functions on $Ω$, $\widetilde{\mathscr{L}}^{(k)}=(-L_1^{(k+2)},L_0^{(k+2)})^{T}$, $\mathscr{L}^{(k)}=(L_0^{(k+1)},L_1^{(k+1)})$, and $L_0^{(k)},L_1^{(k)}$ are multiplication-like operators on $k$-regular functions. This gives the quaternionic analogue of the classical Koszul complex. And we present the long exact sequence in cohomology for the case $Ω\cap\{q_0=0\}=\emptyset$ with explicit differential connecting maps, by applying the Cauchy-Fueter complex and cohomological methods. As an application, in the special case $n=1, k=1$, the operator pair $(L_0^{(1)}, L_1^{(1)})$ is shown to be surjective if and only if $H^3(Ω, \mathbb{R})=0$. Furthermore, a cohomological vanishing criterion is given for $H^1(Ω,\mathcal{I}_X^{(k)})$; under this criterion, every $k$-regular function on $\{q_0=0\}\capΩ$ extends to a $k$-regular function on $Ω$.

math.CV↗

arXiv2026

NowcastDiT: Diffusion Transformers are Effective Precipitation Nowcasters

Precipitation nowcasting demands accurate short-term forecasts under strong spatiotemporal variability. Diffusion models are well suited to modeling complex precipitation distributions, yet existing approaches often introduce increasingly specialized designs, leaving the capability of a standard diffusion architecture underexplored. We show that a standard Diffusion Transformer already provides a simple and scalable foundation for precipitation nowcasting, with domain-specific requirements accommodated naturally within its design space. Based on this principle, we develop NowcastDiT and instantiate this flexibility through two complementary adaptations: a dynamics-aware noise prior for temporally coherent forecasts, and end-to-end reinforcement learning with timestep-aware rewards for meteorological skill. Experiments on SEVIR and MRMS benchmarks show that NowcastDiT achieves state-of-the-art performance in both perceptual quality and meteorological skill. These results suggest that standard DiT can serve as an effective foundation for precipitation nowcasting.

cs.LG↗

arXiv2026

VERPO: Verified Evidence Regularized Policy Optimization

Verifiable rewards improve language models through reliable task-level feedback, but methods based on Group Relative Policy Optimization (GRPO) apply a sequence-level advantage uniformly across all tokens. This coarse credit assignment reinforces or penalizes entire responses without identifying which local decisions to preserve, reinforce, or revise. Conversely, evidence-conditioned self-distillation provides denser token-level supervision, yet teacher imitation can transfer stylistic artifacts and miscalibrated confidence that destabilize training when misaligned with task success. We introduce VERPO, which converts evidence-conditioned guidance into reward-aligned token-level credit assignment while retaining the outcome objective. VERPO decomposes teacher guidance into an evidence-free reference term and signed, evidence-induced corrections at each token. A stopped controller combines selective acceptance, token-wise localization, and cost-aware scaling by balancing alignment with the local GRPO update direction against Fisher movement cost. Furthermore, we introduce Fisher Evidence Contrast (FEC), which attenuates nuisance shifts along an estimated evidence-presence direction through a regularized projection. Across five scientific reasoning and tool-use tasks, VERPO prevents optimization collapse and consistently achieves the highest multi-task average across model backbones, yielding marked improvements particularly on smaller models over strong baselines. Qualitative diagnostics confirm that token acceptance selectively targets reasoning bottlenecks consistent with local reward alignment and Fisher movement cost.

cs.LG↗

arXiv2026

Quantization-Aware EE Optimization and SE-EE Tradeoff for MiLAC-Aided MU-MISO Beamforming

In large antenna arrays, hardware power consumption becomes a dominant design constraint, making energy efficiency (EE) a first-class objective alongside spectral efficiency (SE). Microwave linear analog computer (MiLAC)-aided beamforming, whose front end is a passive reciprocal stream-to-antenna network, addresses this tension by reducing the active radio-frequency chain count to the stream number, at a moderate SE cost. Despite this promise, no EE optimization framework has been established for MiLAC-aided beamforming that accounts for digital-to-analog converter quantization noise and post-quantized transmit power. We fill this gap for downlink multiuser multiple-input single-output (MU-MISO) systems by formulating quantization-aware EE maximization over the MiLAC-feasible beamformer and characterizing the resulting SE-EE tradeoff. Three contributions follow. First, we prove a row-space optimality property of the effective MiLAC-aided beamformer, yielding an equivalent reduced-dimension reformulation whose complexity scales with the stream number rather than the antenna number. Second, we develop a low-complexity Dinkelbach-weighted minimum mean-square error algorithm aided by projected gradient descent that is guaranteed to converge to a stationary point. Third, we cast the SE-EE tradeoff as a multi-objective problem and trace its Pareto boundary via a weighted-sum method that combines an alternative reduced-dimension coordinate with auxiliary-variable successive convex approximation, yielding convex per-iteration subproblems with guaranteed convergence. Numerical results on a DeepMIMO v4 deployment show MiLAC-aided beamforming substantially improves EE over digital and hybrid benchmarks at a moderate SE cost and significantly expands the achievable SE-EE operating region.

eess.SP↗