Searcharxiv⌕ Search

arXiv subjects

Sihan Wang

Publications and source records attributed to Sihan Wang.

At least 19 recordsLinked to original sources
arXiv2026

DeepRadio: A Human-Agent Workspace for Physical Radio Development

Agentic programming can automate software-defined radio (SDR) development, but decisions can propagate from software to hardware without sufficient user visibility or control. We present DeepRadio, a human-agent workspace embedded in GNU Radio Companion (GRC) for jointly constructing, verifying, and operating physical radio systems. DeepRadio uses a dynamic and state-driven workflow to coordinate a traceable radio specification, claim-oriented cross-layer verification, and mixed-initiative control. These mechanisms expose decisions, tie conclusions to available evidence, and return control to users before consequential hardware actions. Across seven atomic SDR tasks, DeepRadio completed nearly all runs while satisfying all prescribed software and radio-frequency constraints. In a 20-participant study, more participants achieved a verified physical deployment with DeepRadio than with Codex-assisted GRC. Participants reported stronger support for shared intent and understanding while retaining initiative and control. These results suggest that DeepRadio can make agent-assisted physical radio development more reliable, understandable, and controllable.

cs.HC↗
arXiv2026

Frequency-Decoupled Diffusion Guidance for Non-Blind Image Deblurring

Pretrained diffusion models provide powerful image priors for training-free posterior sampling in image restoration. To guide this sampling process, frequency-aware methods progressively incorporate measurement information across frequency bands, facilitating coarse-to-fine reconstruction. However, existing methods typically do not explicitly separate frequency activation from degradation-induced attenuation, leaving attenuation differences among inactive frequencies insufficiently modeled. In this work, we propose frequency-decoupled posterior guidance to separate frequency activation from attenuation-aware spectral regularization. Specifically, a progressive low-to-high frequency schedule determines the active measurement band, while a kernel-derived attenuation map defines a selective spectral prior over inactive components. To stabilize the sampling process, we also introduce a local trajectory regularizer that suppresses spatially irregular state-to-clean deviations. For a fixed endpoint energy, we provide a KL-regularized path-space interpretation. In practice, we construct time-dependent guidance through local energy corrections using a Tweedie plug-in approximation. Experiments on natural-image benchmarks demonstrate strong PSNR and SSIM performance across challenging non-blind deblurring settings, even at higher measurement noise levels.

cs.CV↗
arXiv2026

ReWorld-Track: A Recursive Event World Model for Language-Guided Multi-Camera Tracking

Language-guided multi-camera tracking must preserve a target identity across unobserved gaps, where similar candidates and uncertain returns can make early associations unreliable. A wrong match can corrupt the history used to predict later observations and propagate identity errors across subsequent camera handoffs. We propose ReWorld-Track, a recursive event world model that carries association uncertainty into future predictions. Candidate matches and continued waiting define alternative target states, whose posterior probabilities are used to update a persistent recurrent belief. This representation preserves uncertainty about alternative trajectories through successive observations. This belief predicts the next camera, arrival time, and entry region, while appearance and language evidence guide association. By training across successive handoffs, the model learns to retain uncertainty that remains useful for later predictions and identity decisions. ReWorld-Track achieves HOTA scores of 65.19 on CityFlowV2 and 45.36 on MTMMC, with improved identity continuity across repeated handoffs. On MTMMC, its structured posterior update gains 0.50 HOTA points over a similarly sized generic updater and 0.94 points over fixed-moment soft association, raising next-camera accuracy from 86.03% to 87.41% and reducing median arrival-time error from 0.78 s to 0.71 s for subsequent target returns.

cs.CV↗
arXiv2026

High-Dimensional Ultra-Log-Concave Distributions

Ultra-log-concave distributions are ubiquitous in probability, combinatorics, and statistical mechanics and have been studied extensively. In this paper, we introduce a quantitative high-dimensional extension of this notion, called $δ$-ultra-log-concavity, for probability measures on $\mathbb{N}^d$ with downward closed support. When $δ= 1$, this notion coincides with the class studied by Gurvits (2009) via strongly log-concave generating functions, and with the class defined by Anari, Oveis Gharan, and Vinzant (2021) via completely log-concave generating functions; in one dimension, it reduces to classical ultra-log-concavity. We establish several functional inequalities, including a weighted Poincaré inequality, a discrete Brascamp--Lieb inequality, and a weighted Wu-type modified log-Sobolev inequality. Our approach combines integrated Bakry--Émery calculus for a canonical birth-death chain with Poisson stochastic localization, which arises as the time reversal of coordinatewise binomial thinning. We further establish concentration of measure, maximum-entropy principles, and several closure properties for ultra-log-concave measures, and develop applications to queueing models, polymatroids, antiferromagnetic Potts models, and hardcore models. Finally, a lattice scaling limit of the discrete theory yields Poincaré and Brascamp--Lieb inequalities for Laguerre diffusions.

math.PR↗
arXiv2026

Optimal Mixing of Glauber Dynamics for the Sherrington-Kirkpatrick Model at $β< 1/2$

We prove that for every fixed inverse temperature $β< 1 / 2$, with high probability over the disorder, the single-site Glauber dynamics for the $n$-spin Sherrington-Kirkpatrick model mixes from every initial configuration to within total variation distance $\varepsilon$ in $O_β\left(n \log\left(n / \varepsilon\right)\right)$ steps. The bound holds uniformly over all external fields and is optimal up to constants depending only on $β$. The main ingredient is a deterministic criterion for optimal-order Poincaré inequalities in general Ising models, established via the integrated Bakry-Émery criterion together with a new two-spin estimate. A standard application of the localization-scheme framework of Chen and Eldan then upgrades the Poincaré inequality to a modified log-Sobolev inequality, yielding the optimal mixing-time bound. The main ideas underlying the proof of the Poincaré inequality were generated by GPT-5.6 Sol Ultra.

math.PR↗
arXiv2026

FishBack: Pullback Fisher Geometry for Optimal Activation Steering in Transformers

Activation steering has emerged as a lightweight approach for modifying language model behavior without parameter updates, yet existing methods remain brittle: unstable across layers and prone to disturbing behavior unrelated to the target concept. We trace these failures to a hidden assumption shared by widely-used methods such as CAA, ActAdd, and ITI: that the intermediate activation space is Euclidean. We show this assumption is fundamentally flawed. The metric that actually governs how a hidden-state perturbation changes the output is the Fisher information metric of the softmax layer, pulled back to the intermediate layer through the Jacobian of the intervening layers. From it we derive a closed-form steering direction, applied to a hidden state at an intermediate layer, that reaches a target concept change with the least non-target distortion. The framework is sharpest in the early and middle intermediate layers, where the metric is strongly non-Euclidean and geometric correction matters most. We evaluate it on three verb-morphology concepts: third-person-singular, progressive, and past-tense inflection, following standard counterfactual-concept evaluation. On GPT-2 Small, this non-Euclidean geometry is borne out empirically, and our method lowers off-target KL divergence by median factors of 1.4--6.5x against individual steering baselines. On Llama-3-8B and Qwen3-8B, it lowers off-target KL by median factors of 1.8--3.6x against individual baselines at the early and middle layers. These results show that geometric correction retains its advantage on larger models with more complex internal structure.

cs.LG↗
arXiv2026

RadioMaster: Multi-Agent System for Autonomous Radio Signal Generation

Translating user intent into physical radio signals is the last critical step in wireless prototyping. It chains protocol planning, baseband synthesis, and hardware configuration. Large language models and multi-agent systems have reshaped software engineering, raising the question of whether they can solve this problem. Yet current models fail at this task, even when augmented with domain tools. Because the stages run sequentially, an error at any stage propagates downstream, so the end-to-end success rate collapses toward zero even when each stage looks locally competent. We introduce RadioMaster, a fully autonomous multi-agent framework that drives user input to verified emissions transmitted over the air. It rests on three synergistic pillars. RadioWiki grounds generation in domain knowledge to suppress hallucination. RadioAgent decomposes the fragile pipeline into independently executable and locally recoverable stages. RadioEmulator gates deployment behind closed-loop physical-layer verification. We further build RadioBench, the first benchmark for autonomous radio signal generation. Extensive real-world evaluations show that RadioMaster substantially outperforms state-of-the-art baselines in configuration viability and signal fidelity, while reducing configuration time by up to 28x.

cs.MA↗
arXiv2026

Collective Coherent Perfect Absorption in a Synthetic Photon-Phonon Lattice

Coherent perfect absorption (CPA) has emerged as a powerful paradigm for controlling classical and quantum light, and has been demonstrated across a broad range of physical platforms. CPA realized in optomechanics relies on interference between the input field and the mechanically scattered field, but is intrinsically confined to the weak-cooperativity regime, resulting in a narrow absorption bandwidth and the mechanical mode remains thermally occupied. Here, we experimentally demonstrate collective interference-induced CPA in a synthetic photon--phonon lattice. By harnessing cavity-reservoir-mediated interactions among Floquet lattice sites, collective interference shifts the CPA condition deep into the high-cooperativity regime. This enables CPA to coexist with ground-state cooling of the mechanical oscillator, together with a broadened non-Lorentzian absorption lineshape and a singular group-delay response. Our results identify collective interference as a route to quantum-compatible perfect absorption and long-lived quantum storage.

quant-ph↗
arXiv2026

Vision as Unified Multimodal Generation

We formulate computer vision as unified multimodal generation, where heterogeneous visual tasks are expressed in the native text and image generation spaces of a unified multimodal model, without task-specific architectures. Under this formulation, SenseNova-Vision uses natural-language instructions and optional visual prompts to specify tasks, target regions or views, and decoding conventions, and generates responses as text for symbolic outputs, images for dense spatial predictions, or mixed text-and-image outputs for compositional tasks. To support large-scale training, we convert diverse computer vision annotations into instruction-response examples compatible with these generation spaces, resulting in the SenseNova-Vision Corpus, a computer-vision instruction-response corpus spanning text, image, and mixed targets. Starting from an off-the-shelf pretrained unified multimodal model, SenseNova-Vision is trained primarily on this corpus, with auxiliary multimodal data used as a capability-preserving mixture, and requires no task-specific prediction heads or architectural modifications. The resulting model covers a broad range of vision tasks, including detection, OCR, keypoint estimation, segmentation, depth estimation, surface normal prediction, point maps, and camera pose estimation, while supporting language-defined variants that combine category, color, region, and other visual cues. Experiments show that a single unified model can match leading task-specialized systems across structured visual understanding, dense geometric prediction, segmentation, and multi-view visual geometry. These results suggest unified multimodal generation as a scalable route for integrating computer vision capabilities into general-purpose foundation models. The model and corpus are publicly available.

cs.CV↗
arXiv2026

Seeing Before Reasoning: Decoupling Perception and Reasoning for Shortcut-Resilient Multimodal On-Policy Self-Distillation

On-policy self-distillation (OPSD) trains a model on its own rollouts and uses a frozen copy to provide dense token-level targets conditioned on a reference target. This works well for LLM reasoning, but a direct extension to multimodal large language models (MLLMs) can create a shortcut: the privileged target may guide tokens mainly based on the text reference target rather than the image. We propose ViGOS, a visually grounded OPSD framework for MLLM post-training. The student first writes a visual description and then reasons toward the final answer. For valid rollouts, an image-only perception teacher supervises the description, while a privileged reasoning teacher supervises the reasoning and final answer on the same student prefix. A reference teacher is used only for invalid rollouts to recover the output format. Across general vision-language, expert reasoning, visual math, spatial grounding, and visual-language-prior benchmarks, ViGOS keeps the main benefits of OPSD and improves image-grounded behavior in shortcut-prone settings.

cs.LG↗
arXiv2026

Fine-Tuned LLM as a Complementary Predictor Improving Ads System

Recommendation systems power engagement and monetization across feeds, ads, and short-video platforms, but translating the latest advances in Large Language Models into Recommendation Systems (RecSys) gains remains rare, particularly in advertising and production-scale real-world industry setups. Prior real-world LLM successes typically fall into three buckets: (a) generative retrieval that directly predicts the next items for candidate generation, (b) late-stage re-ranking that uses LLMs, and (c) auxiliary signal enrichment with LLMs. We introduce a complementary paradigm for ads: a fine-tuned open-source LLM used not as a ranker, but as an ads-specific ancillary predictor, forecasting likely advertisers from user profiles and histories. This LLM-driven advertiser prediction augments conventional candidate generation and provides informative priors to downstream ranking. Developed in a large-scale production advertising system, our approach produces substantial offline improvements and measurable online business impact, demonstrating that LLM world knowledge and predictive capacity can be efficiently harnessed. Beyond validating LLMs for ads applications, our results show that targeted ancillary predictions can unlock end-to-end gains across both retrieval and late-stage ranking, offering a practical path to LLM-enhanced recommendation at scale.

cs.IR↗
arXiv2026

Discrete Optimal Transport: Rapid Convergence of Simulated Annealing Algorithms

We develop a discrete optimal transport framework for analyzing simulated annealing algorithms on finite state spaces. Building on the discrete Wasserstein metric introduced by Maas (J. Funct. Anal., 2011), we define a generalized discrete Wasserstein-2 distance and the associated notion of \emph{discrete action} for paths of probability measures on graphs. Using these tools, we establish non-asymptotic convergence guarantees for simulated annealing: the KL divergence between the algorithm's output and the target distribution is controlled by the discrete action of the annealing path. This can be viewed as the discrete counterpart of the action-based analysis of annealed Langevin dynamics in continuous spaces by Guo, Tao, and Chen (ICLR 2025). As applications, we analyze simulated annealing for two fundamental models in statistical physics. For the \emph{mean-field Ising model}, we show that annealed single-site Glauber dynamics achieves $\varepsilon$ error in KL divergence in $O(n^5β^2/\varepsilon)$ steps at \emph{any} inverse temperature $β\ge 0$. For the \emph{mean-field $q$-state Potts model}, we show that annealed $(q-1)$-block Glauber dynamics achieves $\varepsilon$ error in $\mathrm{poly}(n, β, 1/\varepsilon)$ steps for all $β\ge β_{\mathsf{s}}=q/2$, the regime where the disordered phase has completely lost stability. In both cases, the key technical contribution is a polynomial upper bound on the discrete action, obtained by exploiting the symmetry of the model to reduce the analysis to a low-dimensional projected chain.

cs.DS↗
arXiv2026

DV-World: Benchmarking Data Visualization Agents in Real-World Scenarios

Real-world data visualization (DV) requires native environmental grounding, cross-platform evolution, and proactive intent alignment. Yet, existing benchmarks often suffer from code-sandbox confinement, single-language creation-only tasks, and assumption of perfect intent. To bridge these gaps, we introduce DV-World, a benchmark of 260 tasks designed to evaluate DV agents across real-world professional lifecycles. DV-World spans three domains: DV-Sheet for native spreadsheet manipulation including chart and dashboard creation as well as diagnostic repair; DV-Evolution for adapting and restructuring reference visual artifacts to fit new data across diverse programming paradigms and DV-Interact for proactive intent alignment with a user simulator that mimics real-world ambiguous requirements. Our hybrid evaluation framework integrates Table-value Alignment for numerical precision and MLLM-as-a-Judge with rubrics for semantic-visual assessment. Experiments reveal that state-of-the-art models achieve less than 50% overall performance, exposing critical deficits in handling the complex challenges of real-world data visualization. DV-World provides a realistic testbed to steer development toward the versatile expertise required in enterprise workflows. Our data and code are available at \href{https://github.com/DA-Open/DV-World}{this project page}.

cs.CL↗
arXiv2026

Pump-Threshold-Free Frequency Comb via Cavity Floquet Engineering

Frequency combs have revolutionized communication, metrology, and spectroscopy. Considerable efforts have been devoted to developing integrated combs, primarily leveraging Pockels or Kerr nonlinearities. Here, we demonstrate an alternative frequency comb generated via cavity Floquet engineering. By periodically modulating the cavity resonance frequency through a driven mechanical oscillator, a Floquet cavity with multiple equally spaced frequency components is created. These sidebands exhibit nearest-neighbor coupling and are phase-locked to the external modulation drive. A pump tone interacts with the pre-modulated cavity to generate the output frequency comb, which we implement in an on-chip microwave cavity optomechanical system. This approach operates independently of a pumping threshold and is insensitive to pump detuning. Consequently, it enables comb generation under far-sideband pumping with nanowatt-scale total power consumption, providing an ultra-low-power platform for integrated frequency comb synthesis.

quant-ph↗
arXiv2025

Evaluating Foundation Models with Pathological Concept Learning for Kidney Cancer

To evaluate the translational capabilities of foundation models, we develop a pathological concept learning approach focused on kidney cancer. By leveraging TNM staging guidelines and pathology reports, we build comprehensive pathological concepts for kidney cancer. Then, we extract deep features from whole slide images using foundation models, construct pathological graphs to capture spatial correlations, and trained graph neural networks to identify these concepts. Finally, we demonstrate the effectiveness of this approach in kidney cancer survival analysis, highlighting its explainability and fairness in identifying low- and high-risk patients. The source code has been released by https://github.com/shangqigao/RadioPath.

cs.AI↗
arXiv2025

Decoupled Entity Representation Learning for Pinterest Ads Ranking

In this paper, we introduce a novel framework following an upstream-downstream paradigm to construct user and item (Pin) embeddings from diverse data sources, which are essential for Pinterest to deliver personalized Pins and ads effectively. Our upstream models are trained on extensive data sources featuring varied signals, utilizing complex architectures to capture intricate relationships between users and Pins on Pinterest. To ensure scalability of the upstream models, entity embeddings are learned, and regularly refreshed, rather than real-time computation, allowing for asynchronous interaction between the upstream and downstream models. These embeddings are then integrated as input features in numerous downstream tasks, including ad retrieval and ranking models for CTR and CVR predictions. We demonstrate that our framework achieves notable performance improvements in both offline and online settings across various downstream tasks. This framework has been deployed in Pinterest's production ad ranking systems, resulting in significant gains in online metrics.

cs.IR↗
arXiv2025

Decoupled Functional Evaluation of Autonomous Driving Models via Feature Map Quality Scoring

End-to-end models are emerging as the mainstream in autonomous driving perception and planning. However, the lack of explicit supervision signals for intermediate functional modules leads to opaque operational mechanisms and limited interpretability, making it challenging for traditional methods to independently evaluate and train these modules. Pioneering in the issue, this study builds upon the feature map-truth representation similarity-based evaluation framework and proposes an independent evaluation method based on Feature Map Convergence Score (FMCS). A Dual-Granularity Dynamic Weighted Scoring System (DG-DWSS) is constructed, formulating a unified quantitative metric - Feature Map Quality Score - to enable comprehensive evaluation of the quality of feature maps generated by functional modules. A CLIP-based Feature Map Quality Evaluation Network (CLIP-FMQE-Net) is further developed, combining feature-truth encoders and quality score prediction heads to enable real-time quality analysis of feature maps generated by functional modules. Experimental results on the NuScenes dataset demonstrate that integrating our evaluation module into the training improves 3D object detection performance, achieving a 3.89 percent gain in NDS. These results verify the effectiveness of our method in enhancing feature representation quality and overall model performance.

cs.CV↗
arXiv2025

Multi-Faceted Large Embedding Tables for Pinterest Ads Ranking

Large embedding tables are indispensable in modern recommendation systems, thanks to their ability to effectively capture and memorize intricate details of interactions among diverse entities. As we explore integrating large embedding tables into Pinterest's ads ranking models, we encountered not only common challenges such as sparsity and scalability, but also several obstacles unique to our context. Notably, our initial attempts to train large embedding tables from scratch resulted in neutral metrics. To tackle this, we introduced a novel multi-faceted pretraining scheme that incorporates multiple pretraining algorithms. This approach greatly enriched the embedding tables and resulted in significant performance improvements. As a result, the multi-faceted large embedding tables bring great performance gain on both the Click-Through Rate (CTR) and Conversion Rate (CVR) domains. Moreover, we designed a CPU-GPU hybrid serving infrastructure to overcome GPU memory limits and elevate the scalability. This framework has been deployed in the Pinterest Ads system and achieved 1.34% online CPC reduction and 2.60% CTR increase with neutral end-to-end latency change.

cs.IR↗