arXiv subjects

Chao Tan

Publications and source records attributed to Chao Tan.

At least 19 recordsLinked to original sources

arXiv2026

Scalable Question-Centric Text-to-Image Evaluation: Reliable Ranking, Fine-Grained Diagnosis, and Cost-Aware Routing

Modern text-to-image (T2I) models often have similar total scores but different strengths, making practical selection difficult. Fine-grained benchmarks decompose prompts into questions, yet often return them to prompt scores and fixed categories, weakening attribution and ignoring complexity. Related requirements are also scored separately or as one total, obscuring basic versus compositional failure. We present QC-T2I-Bench, a question-centric framework that converts open prompts into attributed atomic questions and organizes their dependencies with Davidsonian Scene Graphs (DSGs). We use hierarchy-constrained question aggregation to exclude downstream questions after a prerequisite fails and to prevent simple and complex prompts from receiving the same total weight. We then use the DSG structure to measure joint success within prompts and compare repeated entities across prompts, separating basic realization failures from failures under additional requirements. We evaluate multiple open-source T2I models on English and Chinese prompts. The resulting question-level evidence supports reliable ranking and fine-grained diagnosis: joint completion falls from 80.7\% for components with two capabilities to 37.2\% for those with seven or more. Finally, we reuse the same records for training-free routing; our cost-aware router matches ERNIE's 89.51-point estimate with 21.3\% less GPU-s/MP.

cs.AI↗

arXiv2026

Feature-domain Fourier ptychographic tomography with dark-field illumination

Fourier ptychographic tomography (FPT) is an implementation of intensity diffraction tomography that reconstructs three-dimensional (3D) refractive-index (RI) distributions from angle-varied intensity measurements. The distinctive advantage of FPT emerges when incorporating dark-field illumination, which extends the space-bandwidth product toward gigavoxel-scale volumetric imaging, yet dark-field measurements are highly sensitive to system imperfections and often have low signal-to-noise ratios. Here, we propose feature-domain FPT (FD-FPT), which evaluates data fidelity after feature extraction and is optimized using automatic differentiation. In numerical and experimental tests, FD-FPT resolves structures near the synthetic-aperture cutoff far more reliably than the spatial-domain baseline (SD-FPT). Notably, in a whole-mount Oedogonium specimen, the reticulate chloroplast network and transverse septa were resolved only by FD-FPT. We further demonstrate a 1.81-gigavoxel RI reconstruction of a mouse adrenal gland section across a 1.66 by 1.40 square millimeter field of view, establishing FD-FPT as a practical route to label-free volumetric imaging that combines millimeter-scale coverage with cellular-scale structural contrast.

physics.optics↗

arXiv2026

Is Multimodal Speculative Decoding Ready for Diffusion-Based Parallel Drafting? A Survey and Empirical Diagnosis

Speculative decoding accelerates autoregressive generation by allowing a lightweight drafter to propose future tokens while a target model verifies them in parallel. Its lossless guarantee has motivated a line of work that pushes the drafter itself toward parallel generation. The most recent paradigm is block-parallel generative drafting, including diffusion-based methods such as DFlash and DSpark, achieving up to 3.6x speedup on common daily chatting tasks. While this transition is well studied in text-only LLMs, its applicability to multimodal models remains an open question. Existing multimodal speculative decoding efforts focus on input compression, adapter alignment, candidate coverage, or modality-specific verification; however, block-parallel generative drafting remains largely unexplored. To bridge this gap, this paper combines a modality-centered survey with a cross-architecture empirical study to ask: Is multimodal speculative decoding ready for diffusion-based parallel drafting? In this survey, we systematically analyze a wide spectrum of multimodal models, spanning Vision-Language, Video-Language, Audio, and Vision-Language-Action (VLA) architectures, from the dual perspectives of drafting parallelism and cross-modal information interaction. We introduce a unified taxonomy that isolates drafter-side parallelism from orthogonal design choices such as tree construction and verification strategies. Furthermore, we provide a comprehensive empirical comparison of existing methods under varying degrees of parallelism across standardized multimodal benchmarks, including OCR, VQA, visual reasoning, and image captioning. Finally, we summarize the limitations of current approaches, discuss open challenges, and outline promising future directions for this rapidly evolving field.

cs.AI↗

arXiv2026

OTCache: Optimal Transport for Geometry-Aware Caching in Diffusion Models

We propose OTCache, a training-free framework for accelerating diffusion sampling via caching schedule prediction. Existing graph-based caching methods reduce redundant computation by optimizing shortest-path objectives, but rely on an additive independence assumption, which often breaks down in the low NFE regime. To address this issue, OTCache models caching schedules across inference budgets as a smooth evolution in policy space, inspired by Optimal Transport (OT). The framework consists of three stages: (1) obtaining a high-fidelity \textbf{reference schedule} using a graph-based caching method under a conservative budget; (2) performing a lightweight anchor search under an extreme low-budget setting via Optuna optimization with an end-to-end perceptual objective; and (3) predicting schedules for target budgets via quantile interpolation between the reference and anchor policies using continuous warping representations. Experiments on FLUX.1 [dev], Qwen-Image, and HunyuanVideo show that OTCache achieves 4.5x, 4.7x, and 3.66x acceleration, respectively, while consistently improving generation fidelity over state-of-the-art caching baselines. This work provides a new perspective on accelerating diffusion models through Optimal-Transport-inspired schedule modeling. Code:https://github.com/UnicomAI/OTCache

cs.LG↗

arXiv2026

MediaClaw: Multimodal Intelligent-Agent Platform Technical Report

MediaClaw is a multimodal agent platform built on the OpenClaw ecosystem. Its core design follows a three-layer architecture of unified abstraction, pluginized extension, and workflow orchestration. The system is intended to address practical deployment pain points in AIGC adoption, including fragmented capabilities, heterogeneous interfaces, disconnected production processes, and limited reuse of high-quality production workflows. \system{} abstracts full-category AIGC capabilities into a unified invocation model, uses plugins to support hot-pluggable capability expansion, and uses task-oriented Skills to turn complex production processes into reusable workflow assets. This report focuses on the architectural design philosophy of MediaClaw, the design logic of its core capability model, and the key engineering trade-offs in implementation. It aims to provide reusable practical reference for building multimodal capability platforms.

cs.AI↗

arXiv2026

PaLMR: Towards Faithful Visual Reasoning via Multimodal Process Alignment

Reinforcement learning has recently improved the reasoning ability of Large Language Models and Multimodal LLMs, yet prevailing reward designs emphasise final-answer correctness and consequently tolerate process hallucinations--cases where models reach the right answer while misperceiving visual evidence. We address this process-level misalignment with PaLMR, a framework that aligns not only outcomes but also the reasoning process itself. PaLMR comprises two complementary components: a perception-aligned data layer that constructs process-aware reasoning data with structured pseudo-ground-truths and verifiable visual facts, and a process-aligned optimisation layer that constructs a hierarchical reward fusion scheme with a process-aware scoring function to encourage visually faithful chains-of-thought and improve training stability. Experiments on Qwen2.5-VL-7B show that our approach substantially reduces reasoning hallucinations and improves visual reasoning fidelity, achieving state-of-the-art results on HallusionBench while maintaining strong performance on MMMU, MathVista, and MathVerse. These findings indicate that PaLMR offers a principled and practical route to process-aligned multimodal reasoning, advancing the reliability and interpretability of MLLMs.

cs.CV↗

arXiv2026

Halt the Hallucination: Decoupling Signal and Semantic OOD Detection Based on Cascaded Early Rejection

Efficient and robust Out-of-Distribution (OOD) detection is paramount for safety-critical applications.However, existing methods still execute full-scale inference on low-level statistical noise. This computational mismatch not only incurs resource waste but also induces semantic hallucination, where deep networks forcefully interpret physical anomalies as high-confidence semantic features.To address this, we propose the Cascaded Early Rejection (CER) framework, which realizes hierarchical filtering for anomaly detection via a coarse-to-fine logic.CER comprises two core modules: 1)Structural Energy Sieve (SES), which establishes a non-parametric barrier at the network entry using the Laplacian operator to efficiently intercept physical signal anomalies; and 2) the Semantically-aware Hyperspherical Energy (SHE) detector, which decouples feature magnitude from direction in intermediate layers to identify fine-grained semantic deviations. Experimental results demonstrate that CER not only reduces computational overhead by 32% but also achieves a significant performance leap on the CIFAR-100 benchmark:the average FPR95 drastically decreases from 33.58% to 22.84%, and AUROC improves to 93.97%. Crucially, in real-world scenarios simulating sensor failures, CER exhibits performance far exceeding state-of-the-art methods. As a universal plugin, CER can be seamlessly integrated into various SOTA models to provide performance gains.

cs.CV↗

arXiv2026

Learning Collective Medication Effects via Multi-level Abstraction for Medication Recommendation

Historical prescriptions and selected candidate drugs relevant to the current visit serve as important references for medication recommendation. However, in the absence of explicit intrinsic principles for semantic composition, existing methods treat synergistic drugs as independent entities and fail to capture their collective therapeutic effects, resulting in a mismatch between medication-level references and longitudinal patient representations. In this paper, we propose MSAM, a novel medication recommendation model that bridges the gap via multi-level medication abstraction. The model introduces a multi-head graph reasoning mechanism to organize flat daily medication sets into clinically meaningful semantic units, serving as intermediate abstraction results to propagate features from individual drugs to higher-level representations. Building on these units, MSAM performs two-stage abstraction over historical prescriptions and selected candidates via intra- and inter-level feature propagation across heterogeneous clinical structures, capturing collective therapeutic effects aligned with patient conditions. Experiments on two real-world clinical datasets show that MSAM consistently outperforms state-of-the-art methods, validating the effectiveness of structural medication abstraction for recommendation.

cs.CE↗

arXiv2026

MeanCache: From Instantaneous to Average Velocity for Accelerating Flow Matching Inference

We present MeanCache, a training-free caching framework for efficient Flow Matching inference. Existing caching methods reduce redundant computation but typically rely on instantaneous velocity information (e.g., feature caching), which often leads to severe trajectory deviations and error accumulation under high acceleration ratios. MeanCache introduces an average-velocity perspective: by leveraging cached Jacobian--vector products (JVP) to construct interval average velocities from instantaneous velocities, it effectively mitigates local error accumulation. To further improve cache timing and JVP reuse stability, we develop a trajectory-stability scheduling strategy as a practical tool, employing a Peak-Suppressed Shortest Path under budget constraints to determine the schedule. Experiments on FLUX.1, Qwen-Image, and HunyuanVideo demonstrate that MeanCache achieves 4.12X and 4.56X and 3.59X acceleration, respectively, while consistently outperforming state-of-the-art caching baselines in generation quality. We believe this simple yet effective approach provides a new perspective for Flow Matching inference and will inspire further exploration of stability-driven acceleration in commercial-scale generative models.

cs.LG↗

arXiv2025

A Machine-Learning Approach for Identifying CME-Associated Stellar Flares in TESS Observations

Coronal mass ejections (CMEs) are major drivers of stellar space weather and can strongly influence the habitability of exoplanets. However, compared to the frequent occurrence of white-light flares, confirmed stellar CMEs remain extremely rare. Whether such flares are commonly accompanied by CMEs is a key question for solar-stellar comparative studies. Using Sun-as-a-star soft X-ray flare light curves observed by the GOES XRS 1--8~\AA\ channel, we compiled a sample of 1,766 M-class and larger solar flares and extracted features with both deep convolutional neural networks and manual methods. Five machine-learning classifiers were trained to distinguish eruptive from confined flares, with the random forest model achieving the best performance (true skill statistic; TSS = 0.31). This TSS value indicates that the model possesses a moderate ability to discriminate between eruptive and confined flares. Normalized white-light and GOES XRS flare light curves show broadly consistent temporal evolution, reflecting their shared energy-release history and supporting a probabilistic transfer of the model to white-light flare data. We applied the best-performing RF model to 41,405 TESS-detected flares on FGKM-type main-sequence stars, predicting that approximately 47% of events show CME-like morphological characteristics, with the model-implied intrinsic association fraction lying in the range 35%--60%. Intriguingly, the CME occurrence rate decreases with increasing flare energy, indicating that the most energetic flares may be more strongly confined by overlying magnetic fields. These results provide new insight into flare-CME connections in diverse stellar environments and have important implications for assessing the impact of stellar eruptive activity on exoplanetary atmospheres.

astro-ph.SR↗

arXiv2025

LeMiCa: Lexicographic Minimax Path Caching for Efficient Diffusion-Based Video Generation

We present LeMiCa, a training-free and efficient acceleration framework for diffusion-based video generation. While existing caching strategies primarily focus on reducing local heuristic errors, they often overlook the accumulation of global errors, leading to noticeable content degradation between accelerated and original videos. To address this issue, we formulate cache scheduling as a directed graph with error-weighted edges and introduce a Lexicographic Minimax Path Optimization strategy that explicitly bounds the worst-case path error. This approach substantially improves the consistency of global content and style across generated frames. Extensive experiments on multiple text-to-video benchmarks demonstrate that LeMiCa delivers dual improvements in both inference speed and generation quality. Notably, our method achieves a 2.9x speedup on the Latte model and reaches an LPIPS score of 0.05 on Open-Sora, outperforming prior caching techniques. Importantly, these gains come with minimal perceptual quality degradation, making LeMiCa a robust and generalizable paradigm for accelerating diffusion-based video generation. We believe this approach can serve as a strong foundation for future research on efficient and reliable video synthesis. Our code is available at :https://github.com/UnicomAI/LeMiCa

cs.CV↗

arXiv2025

ELGAR: Expressive Cello Performance Motion Generation for Audio Rendition

The art of instrument performance stands as a vivid manifestation of human creativity and emotion. Nonetheless, generating instrument performance motions is a highly challenging task, as it requires not only capturing intricate movements but also reconstructing the complex dynamics of the performer-instrument interaction. While existing works primarily focus on modeling partial body motions, we propose Expressive ceLlo performance motion Generation for Audio Rendition (ELGAR), a state-of-the-art diffusion-based framework for whole-body fine-grained instrument performance motion generation solely from audio. To emphasize the interactive nature of the instrument performance, we introduce Hand Interactive Contact Loss (HICL) and Bow Interactive Contact Loss (BICL), which effectively guarantee the authenticity of the interplay. Moreover, to better evaluate whether the generated motions align with the semantic context of the music audio, we design novel metrics specifically for string instrument performance motion generation, including finger-contact distance, bow-string distance, and bowing score. Extensive evaluations and ablation studies are conducted to validate the efficacy of the proposed methods. In addition, we put forward a motion generation dataset SPD-GEN, collated and normalized from the MoCap dataset SPD. As demonstrated, ELGAR has shown great potential in generating instrument performance motions with complicated and fast interactions, which will promote further development in areas such as animation, music education, interactive art creation, etc.

cs.GR↗

arXiv2024

Investigating Effective Speaker Property Privacy Protection in Federated Learning for Speech Emotion Recognition

Federated Learning (FL) is a privacy-preserving approach that allows servers to aggregate distributed models transmitted from local clients rather than training on user data. More recently, FL has been applied to Speech Emotion Recognition (SER) for secure human-computer interaction applications. Recent research has found that FL is still vulnerable to inference attacks. To this end, this paper focuses on investigating the security of FL for SER concerning property inference attacks. We propose a novel method to protect the property information in speech data by decomposing various properties in the sound and adding perturbations to these properties. Our experiments show that the proposed method offers better privacy-utility trade-offs than existing methods. The trade-offs enable more effective attack prevention while maintaining similar FL utility levels. This work can guide future work on privacy protection methods in speech processing.

eess.AS↗

arXiv2024

Synthetic Test Data Generation Using Recurrent Neural Networks: A Position Paper

Testing in production-like test environments is an essential part of quality assurance processes in many industries. Provisioning of such test environments, for information-intensive services, involves setting up databases that are rich-enough to enable simulating a wide variety of user scenarios. While production data is perhaps the gold-standard here, many organizations, particularly within the public sectors, are not allowed to use production data for testing purposes due to privacy concerns. The alternatives are to use anonymized data, or synthetically generated data. In this paper, we elaborate on these alternatives and compare them in an industrial context. Further we focus on synthetic data generation and investigate the use of recurrent neural networks for this purpose. In our preliminary experiments, we were able to generate representative and highly accurate data using a recurrent neural network. These results open new research questions that we discuss here, and plan to investigate in our future research.

cs.SE↗

arXiv2024

Graph-Based Multivariate Multiscale Dispersion Entropy: Efficient Implementation and Applications to Real-World Network Data

We introduce Multivariate Multiscale Graph-based Dispersion Entropy (mvDEG), a novel, computationally efficient method for analyzing multivariate time series data in graph and complex network frameworks, and demonstrate its application in real-world data. mvDEG effectively combines temporal dynamics with topological relationships, offering enhanced analysis compared to traditional nonlinear entropy methods. Its efficacy is established through testing on synthetic signals, such as uncorrelated and correlated noise, showcasing its adeptness in discerning various levels of dependency and complexity. The robustness of mvDEG is further validated with real-world datasets, effectively differentiating various two-phase flow regimes and capturing distinct dynamics in weather data analysis. An important advancement of mvDEG is its computational efficiency. Our optimized algorithm displays a computational time that grows linearly with the number of vertices or nodes, in contrast to the exponential growth observed in classical methods. This efficiency is achieved through refined matrix power calculations that exploit matrix and Kronecker product properties, making our method faster than the state of the art. The significant acceleration in computational time positions mvDEG as a transformative tool for extensive and real-time applications, setting a new benchmark in the analysis of time series recorded at distributed locations and opening avenues for innovative applications.

math.CO↗

arXiv2023

Dispersion entropy: A Measure of Irregularity for Graph Signals

We introduce a novel method, called Dispersion Entropy for Graph Signals, $DE_G$, as a powerful tool for analysing the irregularity of signals defined on graphs. We demonstrate the effectiveness of $DE_G$ in detecting changes in the dynamics of signals defined on synthetic and real-world graphs, by defining mixed processing on random geometric graphs or those exhibiting with small-world properties. Remarkably, $DE_G$ generalises the classical dispersion entropy for univariate time series, enabling its application in diverse domains such as image processing, time series analysis, and network analysis, as well as in establishing theoretical relationships (i.e., graph centrality measures, spectrum). Our results indicate that $DE_G$ effectively captures the irregularity of graph signals across various network configurations, successfully differentiating between distinct levels of randomness and connectivity. Consequently, $DE_G$ provides a comprehensive framework for entropy analysis of various data types, enabling new applications of dispersion entropy not previously feasible, and revealing relationships between graph signals and its graph topology.

math.CO↗

arXiv2022

A noise-robust Multivariate Multiscale Permutation Entropy for two-phase flow characterisation

Using a graph-based approach, we propose a multiscale permutation entropy to explore the complexity of multivariate time series over multiple time scales. This multivariate multiscale permutation entropy (MPEG) incorporates the interaction between channels by constructing an underlying graph for each coarse-grained time series and then applying the recent permutation entropy for graph signals. Given the challenge posed by noise in real-world data analysis, we investigate the robustness to noise of MPEG using synthetic time series and demonstrating better performance than similar multivariate entropy metrics. Two-phase flow data is an important industrial process characterised by complex, dynamic behaviour. MPEG characterises the flow behaviour transition of two-phase flow by incorporating information from different scales. The experimental results show that MPEG is sensitive to the dynamic of flow patterns, allowing us to distinguish between different flow patterns.

physics.data-an↗

arXiv2022

Multivariate permutation entropy, a Cartesian graph product approach

Entropy metrics are nonlinear measures to quantify the complexity of time series. Among them, permutation entropy is a common metric due to its robustness and fast computation. Multivariate entropy metrics techniques are needed to analyse data consisting of more than one time series. To this end, we present a multivariate permutation entropy, $MPE_G$, using a graph-based approach. Given a multivariate signal, the algorithm $MPE_G$ involves two main steps: 1) we construct an underlying graph G as the Cartesian product of two graphs G1 and G2, where G1 preserves temporal information of each times series together with G2 that models the relations between different channels, and 2) we consider the multivariate signal as samples defined on the regular graph G and apply the recently introduced permutation entropy for graphs. Our graph-based approach gives the flexibility to consider diverse types of cross channel relationships and signals, and it overcomes with the limitations of current multivariate permutation entropy.

math.CO↗