arXiv subjects

Mingyu Zhao

Publications and source records attributed to Mingyu Zhao.

At least 19 recordsLinked to original sources

arXiv2026

Metacognitive Decoding: Efficient Inference for Diffusion Language Agents

Diffusion language models offer a promising path to efficient agents through parallel token generation. However, efficient interaction requires adapting both where computation is spent and how much text is generated. We introduce a metacognitive controller that guides diffusion generation according to the needs of the current response and the agent's interaction with the environment. It learns when to finalize tokens and how far generation should extend, reducing overall computation within responses. Our controller, a lightweight plug-in for the diffusion backbone, is trained on agent trajectories to balance correctness and generation cost. This approach connects the agent's generation process to the interaction the agent serves. Experiments across five benchmarks demonstrate that our approach reduces inference FLOPs by 20.6% on average compared with the most efficient baseline, while maintaining competitive task accuracy. Ablation studies further validate the effectiveness of our approach.

cs.CL↗

arXiv2026

AMBER: Training Long-Horizon Web Agents through Append-Only Memory

Modern language-model agents increasingly interact with external environments over long-horizon, multi-step trajectories, where the accumulated interaction history can quickly exceed practical context budgets. To ensure reliability, agents must maintain factual information over long horizons, remember execution errors and corrective feedback, and track progress across actions. Several approaches have been proposed to achieve this without the need for maintaining the entire execution history in context, such as using the reasoning and action history, learning to maintain a fixed-size memory through an overwrite mechanism, and periodic summarization. Although overwrite memory can in principle retain anything an append-only memory can, it must learn to carry each fact through every subsequent rewrite, which is difficult to learn from sparse outcome rewards; for interactive applications like web agents, we find that trained overwrite memories delete key information required by the trajectory, as well as corrective feedback received from the environment. We introduce AMBER (Append-only Memory Bank for Evidence Retention) - a simple and scalable framework where an agent jointly learns to reason, act, and write free-form memory, while an append-only rule guarantees retention by construction. This allows AMBER to be trained end-to-end with reinforcement learning from outcome rewards without the need for extensive curated SFT data. On WebArena Lite, AMBER improves average success over overwrite-based memory by 4.09 percentage points, increases the fraction of tasks solved in five repeated runs by 4.8 percentage points, and matches an overwrite baseline trained on substantially more expensive curated supervision. AMBER achieves these improvements while maintaining a practical token budget, providing a strong balance between context efficiency, task performance, and reliable long-horizon execution.

cs.AI↗

arXiv2026

SimTrace: Grounded Multimodal User Trajectories Generation for Online User Modeling

Virtual clients offer a cost-effective approach to support applications such as A/B testing, recommender system development, and interface evaluation. However, building them requires access to large-scale, semantically faithful, fine-grained online user trajectories. These data are difficult to obtain because proprietary logs are subject to privacy restrictions and small businesses often lack sufficient traffic. Consequently, existing public datasets either abstract away fine-grained user interaction details or preserve rich context but remain platform-specific and small-scale. To address this gap, we propose SimTrace, a framework that generates faithful, fine-grained synthetic multimodal clickstreams through a computer-use client agent that is grounded in real user trajectories and the given web environment. SimTrace anonymizes real interactions and constructs a simulated twin of the given web environment, then uses both to generate synthetic interaction trajectories. Each action is paired with its corresponding web observations and user context, yielding a shareable alternative to confidential logs for developing computer-use agent-style virtual clients. We apply SimTrace to an e-commerce setting and evaluate both its fidelity and downstream utility. SimTrace outperforms competing baselines on 7 out of 8 fidelity metrics. Models trained on synthetic data achieve performance comparable to those trained on real data on downstream tasks such as purchase prediction and recommendation. For next action prediction task, augmenting real data with synthetic data further improves accuracy by 11.0% relative to training on real data alone. We release SimTrace as an open-source package to facilitate research on online user behavior modeling.

cs.AI↗

arXiv2026

Audio Tokens as a Budgeted Resource: Marginal-Utility Allocation for Scalable Audio Representations

Discrete audio tokens are widely used as a representation interface, yet fixed-depth RVQ tokenizers allocate equal capacity to every frame despite varying refinement value. We introduce UniAdapt, which learns marginal utility of RVQ refinements on a frozen codec and allocates them under exact serialized-bit budgets. A rate-independent causal controller predicts acoustic utility, while an optional semantic head supports speech-only utterance-level allocation; measured acoustic and semantic marginal gains on speech have a correlation of 0.42. For causal allocation, a primal-dual allocator selects prefix-valid depths, while an exact guard constrains each sequence prefix to its matched fixed-depth serialized budget. Under utterance-level allocation, UniAdapt reduces Log-STFT distortion by 1.07-4.39 percent across speech, music, and environmental audio without larger budgets. Causally, it improves three of four speech rates with zero violations across 800 utterance-rate evaluations and runs faster than real time. A 20-listener utterance-level MUSHRA study shows a significant 3.52-point speech improvement, with no significant differences on music or environmental audio. These results support separating utility prediction from budget enforcement for scalable, budget-conditioned audio representations.

eess.AS↗

arXiv2026

UniStream: Multi-Expert Residual Vector Quantization for 48 kHz Causal Streaming Audio Coding

We present UniStream, a fully causal 48 kHz neural audio codec for streaming speech, music, and environmental sounds. At its core is Multi-Expert Residual Vector Quantization (ME-RVQ), which replaces the single shared codebook in each residual quantization layer with four expert codebooks controlled by a deterministic Top-K router. Because routing decisions are derived solely from previously decoded quantized states, the decoder can reproduce the selected experts without transmitting expert identifiers, thereby expanding quantization capacity while adding 5.5M parameters. We further introduce an auxiliary Optimal Transport Conditional Flow Matching (OT-CFM) objective to regularize the quantized latent space during training. The flow module is removed entirely at inference and therefore incurs no runtime overhead. UniStream supports a 12 kbps Top-1 mode and a 22.5 kbps Top-2 mode within a causal 48 kHz encoder-decoder framework, while achieving real-time GPU inference. To complement narrow-band speech metrics, we report 48 kHz ViSQOL audio mode, ViSQOL speech mode, standard VGGish-FAD, DNSMOS P.835, and higher-rate reference comparisons with Opus and EnCodec. At 12 kbps, UniStream-Top1 achieves PESQ and UTMOS scores comparable to EnCodec while reducing speech Mel-D from 13.07 to 8.21. At 22.5 kbps, UniStream-Top2 achieves a ViSQOL speech-mode score of 4.67 and an environmental audio-mode score of 3.96, exceeding all evaluated systems operating at 12 kbps or below in the latter setting. It also comes within 0.03 MOS-LQO of Opus at 24 kbps on speech in ViSQOL audio mode. Ablation studies confirm that ME-RVQ is the primary source of quality improvement, whereas OT-CFM provides perceptual gains on speech with a mild trade-off in spectral distortion.

eess.AS↗

arXiv2026

BAMU: Bitstream-Aware Marginal-Utility Allocation for Frozen Pretrained Neural Speech Codecs

Pretrained neural speech codecs typically use a fixed residual vector quantization (RVQ) depth for all frames, ignoring temporal variation in quantization difficulty. We propose BAMU, a bitstream-aware dynamic RVQ allocation framework for frozen pretrained codecs. A lightweight, rate-independent predictor estimates frame- and layer-wise marginal latent-distortion reductions, while a constrained allocator selects prefix-valid depths under an exact serialized-size budget. Experiments on EnCodec and DAC over LibriSpeech, together with VCTK evaluation, show consistent EnCodec gains and DAC improvements mainly at medium and high rates. A 30-listener study confirms a MOS improvement from 3.449 to 3.780 over matched fixed-depth coding.

eess.AS↗

arXiv2026

A Robust EDM Optimization Approach for 3D Single-Source Localization with Angle and Range Measurements

Accurate source localization in Multi-Platform Radar Networks (MPRNs) benefits from exploiting both range and angle measurements under robust estimation. In this paper, we propose a robust Euclidean distance matrix (EDM) optimization model that simultaneously integrates range measurements, angle information, and the least absolute deviation ($\ell_1$-norm) criterion for the case of 3D single-source localization (3DSSL). A key theoretical contribution of this work is the rigorous reformulation of {existing} 3D angle measurements into simple box constraints on the Euclidean distances. Unlike previous approximations, we achieve this by reducing each of the 3D angle measurements to a two-dimensional nonlinear optimization problem, whose global minimum and maximum solutions can be characterized and utilized to get the lower and upper bounds of the distances from the unknown source to the sensors. To solve the resulting rank-constrained EDM problem, we develop an efficient algorithm based on the majorization penalty method. Extensive numerical experiments confirm that the new EDM model significantly outperforms leading solvers in terms of localization accuracy and computational efficiency, particularly in low Signal-to-Noise Ratio (SNR) scenarios.

eess.SP↗

arXiv2026

PIT-SUN: A Deployable Empirical Marginal Transform Framework with Expectation-Consistent Recovery for Regression in Recommender Systems

Estimating original-space conditional expectations is central to value-driven recommender systems, including dwell time, GMV, and LTV forecasting. Standard MSE is expectation-consistent in principle, but its gradients become unstable on heavy-tailed, zero-inflated, and multimodal targets, causing mean collapse and tail shrinkage. Target transformation alleviates this scale conflict, yet any useful nonlinear marginal transform loses expectation consistency under direct inversion. This is not an implementation oversight: a direct inverse-transform estimator is universally expectation-consistent only when the inverse transform is affine, which cannot simultaneously provide bounded tail compression. Existing conditionally linear recovery methods restore expectation consistency, but still leave open which coordinate, inverse lookup, recovery base, and deployment monitor should be selected for sparse complex marginals. We propose \textbf{P}robability-\textbf{I}ntegral-\textbf{TranS}formed \textbf{Un}biased recovery (\textbf{PIT-SUN}), a deployable empirical marginal recovery framework. PIT-SUN uses one empirical marginal table to define a bounded normal-score coordinate, its inverse-quantile lookup, a variance-controlled recovery base, and drift monitoring, then applies multiplicative SUN recovery to estimate the original-space expectation instead of directly inverting transformed predictions. Experiments on synthetic distributions, public benchmarks, large-scale industrial datasets, and online deployment show robust improvements in point accuracy, calibration, and ranking quality with lightweight deployment overhead.

cs.LG↗

arXiv2026

Research on the Flat Field Measurement Method of Coronagraph

The solar corona has an extremely low density, and its brightness is only about one millionth of that of the photosphere. High-dynamic-range imaging of its faint structure is therefore essential for studying coronal heating, coronal mass ejections, and space weather. Quantitative coronagraph imaging requires flat-field measurement and calibration, which underpin intensity calibration, small-scale feature detection, and long-term cyclic analysis. This paper analyzes the coronagraph imaging chain and the origins of flat-field errors, including optical aberrations, stray light, and pixel-response non-uniformity, and summarizes the resulting calibration requirements of next-generation coronagraphs. On this basis, ground-based and space-based flat-fielding methods are systematically reviewed: the ground-based methods include integrating-sphere uniform light sources, opal glass/diffuser plates, clear-sky and thin-cloud backgrounds, and solar-disk scanning, while the space-based methods include internal light sources and diffuser plates, attitude-roll and off-corona offset observations, and multi-phase statistical self-consistent flat-fielding. Their accuracy, resource cost, and applicability are compared. The review shows that no single method is simultaneously high-precision, easy to update, and engineer-friendly; a hierarchical, multi-method calibration framework is therefore recommended. Finally, a new method is proposed in which lithographically generated structured light fields, combined with Fourier-optics and machine-learning inversion, are used to estimate the pixel-response function. Preliminary experiments show that this method achieves a lower residual error than the integrating-sphere and opal-glass methods, providing a high-precision reference for future wide-band, high-resolution coronagraph calibration.

astro-ph.SR↗

arXiv2026

Observational Technological Innovations and Future Development of the Lijiang Coronagraph

As a core ground-based coronal observation facility in China's low-latitude high-altitude regions, the Lijiang Coronagraph leverages the natural advantages of Lijiang Astronomical Observation Station, including its 3200 m altitude and low atmospheric turbulence. It has undergone a full development process, from introduction via Chinese-Japanese cooperation to independent innovation and iteration. This paper systematically summarizes its core technological innovations: upgrade of the automatic operating system, integration of the dual-band observation system, stray light suppression based on image differencing before and after cleaning, and high-precision image calibration and registration. These advances have significantly improved observation efficiency and data quality, laying a solid foundation for high-quality observations. Scientifically, the data reveal that 1.1 solar radii is a highly correlated region between coronal green line brightness and magnetic field intensity. The study also confirms a strong correlation between the coronal green line and the SDO/AIA 21.1 nm extreme ultraviolet band (correlation coefficient: 0.89-0.99), supporting early warning research on Coronal Mass Ejections (CMEs). These results provide key data for verifying coronal heating mechanisms and exploring the origin of the slow solar wind. The experience from the Lijiang Coronagraph not only lays a foundation for China's next-generation large-aperture coronagraphs, but also accelerates progress in low coronal observation capabilities, enabling the country to build internationally competitive capabilities in this field. The system is also an important part of the global coronal observation network.

astro-ph.SR↗

arXiv2026

SimGym: A Framework for A/B Test Simulation in E-Commerce with Traffic-Grounded VLM Agents

A/B testing remains the gold standard for evaluating modifications to e-commerce storefronts, yet it diverts traffic, requires weeks to reach statistical significance, and risks degrading user experience. We present SimGym, a framework for simulating A/B tests on e-commerce storefronts using vision-language model (VLM) agents operating in a live browser. The framework comprises three key components: (a) a traffic-grounded persona generation pipeline that derives per-shop buyer archetypes and intents from production clickstream data; (b) a live-browser agent architecture that combines multimodal perception over visual and browser-structured observations with episodic memory and guardrails to conduct coherent shopping sessions across control and treatment storefronts; and (c) an evaluation protocol that compares simulated outcome shifts with observed shifts in real buyer behavior. We validate SimGym on A/B tests of visually driven UI theme changes from a major e-commerce platform across diverse storefronts and product categories. Empirical results show that SimGym agents achieve strong agreement with observed outcome shifts, attaining 77% directional alignment with add-to-cart shifts observed across interface variants in real-buyer traffic. It reduces experimental cycles from weeks to under an hour, enabling rapid experimentation without exposing real buyers to candidate variants.

cs.AI↗

arXiv2026

ShopGym: An Integrated Framework for Realistic Simulation and Scalable Benchmarking of E-Commerce Web Agents

Developing and evaluating e-commerce web agents requires environments that preserve meaningful task structure while enabling controllable, reproducible, and scalable scientific comparison. Existing methodologies force a tradeoff: live storefronts provide realism but are non-stationary, difficult to inspect, and irreproducible, while hand-built sandbox benchmarks provide control but cover only a narrow range of layouts, catalogs, policies, and interaction patterns. We argue that the core bottleneck is methodological: the field lacks a scalable way to construct evaluation settings that are simultaneously realistic, diverse, controllable, inspectable, and reproducible. We introduce ShopGym, an integrated framework for realistic simulation and scalable benchmarking of e-commerce web agents. ShopGym is a framework for constructing e-commerce simulation environments and grounded benchmark tasks. Its simulation layer, ShopArena, converts live seed storefronts into self-contained sandbox shops through anonymized shop specifications and a staged, validated generation process. On top of these simulated storefronts, ShopGuru synthesizes benchmark tasks across seven skill categories, grounding each task in the shop's catalog, navigation structure, policies, and interaction affordances. Together, ShopArena and ShopGuru produce self-contained, resettable, inspectable, and stable evaluation artifacts that preserve structural properties and agent-evaluation signals relevant to shopping tasks. We validate the framework through graph-based structural analysis and agent-based behavioral evaluation with 224 generated tasks across six sandbox shops: three constructed with synthetic data and three with real data. Our results show that the synthetic shops preserve key structural properties of live storefronts, with agent performance on synthetic shops positively correlated with performance on live storefronts.

cs.AI↗

arXiv2026

Forward Modeling of Dust-Induced Stray Light in Ground-Based Coronagraphs: A Dual-Path Monitoring Approach for High-Precision Inner Corona Observations

High-precision ground-based observations of the inner corona (1.05-2.0 R_sun) are fundamentally constrained by instrumental stray light, particularly the additive background from dynamic dust accumulation on the objective lens. To address this issue, we propose a correction method for the Spectral Imaging Coronagraph (SICG) based on dual-path real-time monitoring and forward physical modeling. By simultaneously imaging the objective lens surface, we obtain deterministic prior information on dust distribution. We construct a physical point-spread function using optical defocus parameters and reconstruct the nonuniform scattering background via convolution. Model parameters are retrieved through data-driven inversion constrained by polar coronal holes. The method demonstrates excellent robustness under varying contamination conditions. After correction, the rms noise in the polar background is reduced by approximately 67% on average, and the signal-to-background ratio improves by a factor of up to 3.7 under heavy contamination conditions. Comparisons with space-based Solar Dynamics Observatory/Atmospheric Imaging Assembly observations indicate that the corrected images recover the morphological structures of streamers with high fidelity. Further radial intensity analysis reveals that the correction process successfully restores the hydrostatic exponential decay characteristic of inner coronal radiation. The fitted decay coefficient corresponds to a plasma temperature of approximately 2.0 MK, consistent with the characteristic formation temperature of the Fe XIV 530.3 nm line. These results demonstrate that the method effectively eliminates the dominant systematic bias in ground-based observations, providing a reliable data foundation for high-precision coronal thermodynamic and dynamic research with the SICG.

astro-ph.SR↗

arXiv2026

Atmospheric turbulence profiling with the Multistar Turbulence Monitor

Accurate characterization of atmospheric optical turbulence is essential for evaluating astronomical sites and optimizing adaptive optics systems. The Multistar Turbulence Monitor (MTM) infers the vertical distribution of the refractive-index structure constant Cn2(z) from differential image motion measured between multiple stellar pairs in short-exposure frames. We present a comprehensive investigation of the MTM method, combining theoretical analysis, instrument-performance assessment, numerical simulations, and on-sky observations obtained at the Daocheng Astronomical Site. Simulations based on a standard HV turbulence model demonstrate that the inversion pipeline robustly recovers both the integrated seeing and the vertical turbulence profile under realistic centroiding noise and varying pixel scales. The Markov Chain Monte Carlo (MCMC) inversion achieves stable results with thirteen discrete height nodes and provides reliable uncertainties. Three nights of MTM measurements at the Daocheng Astronomical Site show that MTM-derived seeing closely tracks simultaneous Differential Image Motion Monitor (DIMM) results, accurately reproducing both short-term fluctuations and nightly averages. These results confirm that MTM provides a simple, portable, and versatile solution for atmospheric turbulence profiling and routine seeing monitoring.

astro-ph.IM↗

arXiv2026

SPG-Codec: Exploring the Role and Boundaries of Semantic Priors in Ultra-Low-Bitrate Neural Speech Coding

Conventional neural speech codecs suffer from severe intelligibility degradation at ultra-low bitrates, where the bottleneck transitions from acoustic distortion to semantic loss. To address this issue, this paper conducts a systematic investigation into the role and fundamental limits of integrating frozen semantic priors -- specifically HuBERT and Whisper -- into neural speech coding. We introduce and quantitatively validate a novel Semantic Retirement phenomenon: while semantic constraints reduce the Word Error Rate (WER) by up to ~10% relatively at 1.5 kbps, their benefits rapidly diminish beyond 6 kbps, indicating a practical capacity boundary. We further uncover a clear trade-off between different prior types: acoustic-rich priors (HuBERT) better preserve prosodic and timbral details, whereas high-level linguistic priors (Whisper) effectively suppress phonetic hallucinations in noisy environments (reducing hallucination rates by 26 percent) and substantially narrow the generalization gap for unseen speakers. Building on these findings, we propose a bitrate-aware regulation strategy that dynamically adjusts prior strength to optimize the trade-off between semantic consistency and perceptual naturalness. Extensive experimental evaluations confirm that our approach achieves competitive intelligibility and noise robustness compared to existing baselines, offering a principled pathway toward ultra-low-bitrate generative speech coding.

eess.AS↗

arXiv2026

M^3-Bench: Multi-Modal, Multi-Hop, Multi-Threaded Tool-Using MLLM Agent Benchmark

We present M^3-Bench, the first benchmark for evaluating multimodal tool use under the Model Context Protocol. The benchmark targets realistic, multi-hop and multi-threaded workflows that require visual grounding and textual reasoning, cross-tool dependencies, and persistence of intermediate resources across steps. We introduce a similarity-driven alignment that serializes each tool call, embeds signatures with a sentence encoder, and performs similarity-bucketed Hungarian matching to obtain auditable one-to-one correspondences. On top of this alignment, we report interpretable metrics that decouple semantic fidelity from workflow consistency. The benchmark spans 28 servers with 231 tools, and provides standardized trajectories curated through an Executor & Judge pipeline with human verification; an auxiliary four large language models (LLMs) judge ensemble reports end-task Task Completion and information grounding. Evaluations of representative state-of-the-art Multimodal LLMs (MLLMs) reveal persistent gaps in multimodal MCP tool use, particularly in argument fidelity and structure consistency, underscoring the need for methods that jointly reason over images, text, and tool graphs. Our Benchmark's anonymous repository is at https://github.com/EtaYang10th/Open-M3-Bench

cs.AI↗

arXiv2026

SimGym: Traffic-Grounded Browser Agents for Offline A/B Testing in E-Commerce

A/B testing remains the gold standard for evaluating e-commerce UI changes, yet it diverts traffic, takes weeks to reach significance, and risks harming user experience. We introduce SimGym, a scalable system for rapid offline A/B testing using traffic-grounded synthetic buyers powered by Large Language Model agents operating in a live browser. SimGym extracts per-shop buyer profiles and intents from production interaction data, identifies distinct behavioral archetypes, and simulates cohort-weighted sessions across control and treatment storefronts. We validate SimGym against real human outcomes from real UI changes on a major e-commerce platform under confounder control. Even without alignment post training, SimGym agents achieve state of the art alignment with observed outcome shifts and reduces experiment cycles from weeks to under an hour , enabling rapid experimentation without exposure to real buyers.

cs.AI↗

arXiv2026

CC-OR-Net: A Unified Framework for LTV Prediction through Structural Decoupling

Customer Lifetime Value (LTV) prediction, a central problem in modern marketing, is characterized by a unique zero-inflated and long-tail data distribution. This distribution presents two fundamental challenges: (1) the vast majority of low-to-medium value users numerically overwhelm the small but critically important segment of high-value "whale" users, and (2) significant value heterogeneity exists even within the low-to-medium value user base. Common approaches either rely on rigid statistical assumptions or attempt to decouple ranking and regression using ordered buckets; however, they often enforce ordinality through loss-based constraints rather than inherent architectural design, failing to balance global accuracy with high-value precision. To address this gap, we propose \textbf{C}onditional \textbf{C}ascaded \textbf{O}rdinal-\textbf{R}esidual Networks \textbf{(CC-OR-Net)}, a novel unified framework that achieves a more robust decoupling through \textbf{structural decomposition}, where ranking is architecturally guaranteed. CC-OR-Net integrates three specialized components: a \textit{structural ordinal decomposition module} for robust ranking, an \textit{intra-bucket residual module} for fine-grained regression, and a \textit{targeted high-value augmentation module} for precision on top-tier users. Evaluated on real-world datasets with over 300M users, CC-OR-Net achieves a superior trade-off across all key business metrics, outperforming state-of-the-art methods in creating a holistic and commercially valuable LTV prediction solution.

cs.LG↗