arXiv subjects

Yiran Wang

Publications and source records attributed to Yiran Wang.

At least 55 records · Page 3Linked to original sources

arXiv2026

MarsRetrieval: Benchmarking Vision-Language Models for Planetary-Scale Geospatial Retrieval on Mars

Data-driven approaches like deep learning are rapidly advancing planetary science, particularly in Mars exploration. Despite recent progress, most existing benchmarks remain confined to closed-set supervised visual tasks and do not support text-guided retrieval for geospatial discovery. We introduce MarsRetrieval, a retrieval benchmark for evaluating vision-language models for Martian geospatial discovery. MarsRetrieval includes three tasks: (1) paired image-text retrieval, (2) landform retrieval, and (3) global geo-localization, covering multiple spatial scales and diverse geomorphic origins. We propose a unified retrieval-centric protocol to benchmark multimodal embedding architectures, including contrastive dual-tower encoders and generative vision-language models. Our evaluation shows MarsRetrieval is challenging: even strong foundation models often fail to capture domain-specific geomorphic distinctions. We further show that domain-specific fine-tuning is critical for generalizable geospatial discovery in planetary settings. Our code is available at https://github.com/ml-stat-Sustech/MarsRetrieval

cs.CV↗

arXiv2026

Natural Language-Driven Global Mapping of Martian Landforms

Planetary surfaces are typically analyzed using high-level semantic concepts in natural language, yet vast orbital image archives remain organized at the pixel level. This mismatch limits scalable, open-ended exploration of planetary surfaces. Here we present MarScope, a planetary-scale vision-language framework enabling natural language-driven, label-free mapping of Martian landforms. MarScope aligns planetary images and text in a shared semantic space, trained on over 200,000 curated image-text pairs. This framework transforms global geomorphic mapping on Mars by replacing pre-defined classifications with flexible semantic retrieval, enabling arbitrary user queries across the entire planet in 5 seconds with F1 scores up to 0.978. Applications further show that it extends beyond morphological classification to facilitate process-oriented analysis and similarity-based geomorphological mapping at a planetary scale. MarScope establishes a new paradigm where natural language serves as a direct interface for scientific discovery over massive geospatial datasets.

cs.AI↗

arXiv2026

Apollo: Unified Multi-Task Audio-Video Joint Generation

Audio-video joint generation has progressed rapidly, yet substantial challenges still remain. Non-commercial approaches still suffer audio-visual asynchrony, poor lip-speech alignment, and unimodal degradation, which can be stemmed from weak audio-visual correspondence modeling, limited generalization, and scarce high-quality dense-caption data. To address these issues, we introduce Apollo and delve into three axes--model architecture, training strategy, and data curation. Architecturally, we adopt a single-tower design with unified DiT blocks and an Omni-Full Attention mechanism, achieving tight audio-visual alignment and strong scalability. Training-wise, we adopt a progressive multitask regime--random modality masking to joint optimization across tasks, and a multistage curriculum, yielding robust representations, strengthening A-V aligned world knowledge, and preventing unimodal collapse. For datasets, we present the first large-scale audio-video dataset with dense captions, and introduce a novel automated data-construction pipeline which annotates and filters millions of diverse, high-quality, strictly aligned audio-video-caption triplets. Building on this, Apollo scales to large datasets, delivering high-fidelity, semantically and temporally aligned, instruction-following generation in both joint and unimodal settings while generalizing robustly to out-of-distribution scenarios. Across tasks, it substantially outperforms prior methods by a large margin and achieves performance comparable to Veo 3, offering a unified, scalable path toward next-generation audio-video synthesis.

cs.CV↗

arXiv2026

MacVQA: Adaptive Memory Allocation and Global Noise Filtering for Continual Visual Question Answering

Visual Question Answering (VQA) requires models to reason over multimodal information, combining visual and textual data. With the development of continual learning, significant progress has been made in retaining knowledge and adapting to new information in the VQA domain. However, current methods often struggle with balancing knowledge retention, adaptation, and robust feature representation. To address these challenges, we propose a novel framework with adaptive memory allocation and global noise filtering called MacVQA for visual question answering. MacVQA fuses visual and question information while filtering noise to ensure robust representations, and employs prototype-based memory allocation to optimize feature quality and memory usage. These designs enable MacVQA to balance knowledge acquisition, retention, and compositional generalization in continual VQA learning. Experiments on ten continual VQA tasks show that MacVQA outperforms existing baselines, achieving 43.38% average accuracy and 2.32% average forgetting on standard tasks, and 42.53% average accuracy and 3.60% average forgetting on novel composition tasks.

cs.CV↗

arXiv2026

SafeMo: Linguistically Grounded Unlearning for Trustworthy Text-to-Motion Generation

Text-to-motion (T2M) generation with diffusion backbones achieves strong realism and alignment. Safety concerns in T2M methods have been raised in recent years; existing methods replace discrete VQ-VAE codebook entries to steer the model away from unsafe behaviors. However, discrete codebook replacement-based methods have two critical flaws: firstly, replacing codebook entries which are reused by benign prompts leads to drifts on everyday tasks, degrading the model's benign performance; secondly, discrete token-based methods introduce quantization and smoothness loss, resulting in artifacts and jerky transitions. Moreover, existing text-to-motion datasets naturally contain unsafe intents and corresponding motions, making them unsuitable for safety-driven machine learning. To address these challenges, we propose SafeMo, a trustworthy motion generative framework integrating Minimal Motion Unlearning (MMU), a two-stage machine unlearning strategy, enabling safe human motion generation in continuous space, preserving continuous kinematics without codebook loss and delivering strong safety-utility trade-offs compared to current baselines. Additionally, we present the first safe text-to-motion dataset SafeMoVAE-29K integrating rewritten safe text prompts and continuous refined motion for trustworthy human motion unlearning. Built upon DiP, SafeMo efficiently generates safe human motions with natural transitions. Experiments demonstrate effective unlearning performance of SafeMo by showing strengthened forgetting on unsafe prompts, reaching 2.5x and 14.4x higher forget-set FID on HumanML3D and Motion-X respectively, compared to the previous SOTA human motion unlearning method LCR, with benign performance on safe prompts being better or comparable. Code: https://github.com/AIGeeksGroup/SafeMo. Website: https://aigeeksgroup.github.io/SafeMo.

cs.CV↗

arXiv2025

HY-Motion 1.0: Scaling Flow Matching Models for Text-To-Motion Generation

We present HY-Motion 1.0, a series of state-of-the-art, large-scale, motion generation models capable of generating 3D human motions from textual descriptions. HY-Motion 1.0 represents the first successful attempt to scale up Diffusion Transformer (DiT)-based flow matching models to the billion-parameter scale within the motion generation domain, delivering instruction-following capabilities that significantly outperform current open-source benchmarks. Uniquely, we introduce a comprehensive, full-stage training paradigm -- including large-scale pretraining on over 3,000 hours of motion data, high-quality fine-tuning on 400 hours of curated data, and reinforcement learning from both human feedback and reward models -- to ensure precise alignment with the text instruction and high motion quality. This framework is supported by our meticulous data processing pipeline, which performs rigorous motion cleaning and captioning. Consequently, our model achieves the most extensive coverage, spanning over 200 motion categories across 6 major classes. We release HY-Motion 1.0 to the open-source community to foster future research and accelerate the transition of 3D human motion generation models towards commercial maturity.

cs.CV↗

arXiv2025

Incorporating Tissue Composition Information in Total-Body PET Metabolic Quantification of Bone Marrow through Dual-Energy CT

Bone marrow (BM) metabolic quantification with 18F-fluorodeoxyglucose (FDG) positron emission tomography (PET) is of broad clinical significance for accurate assessment of BM at staging and follow-up, especially when immunotherapy is involved. However, current methods of quantifying BM may be inaccurate because the volume defined to measure bone marrow may also consist of a fraction of trabecular bone in which 18F-FDG activity is negligible, resulting in a potential underestimation of true BM uptake. In this study, we demonstrate this bone-led tissue composition effect and propose a bone fraction correction (BFC) method using X-ray dual-energy computed tomography (DECT) material decomposition. This study included ten scans from five cancer patients who underwent baseline and follow-up dynamic 18F-FDG PET and DECT scans using the uEXPLORER total-body PET/CT system. The voxel-wise bone volume fraction was estimated from DECT and then incorporated into the PET measurement formulas for BFC. The standardized uptake value (SUV), 18F-FDG delivery rate K1, and net influx rate Ki values in BM regions were estimated with and without BFC and compared using the statistical analysis. The results first demonstrated the feasibility of performing voxel-wise material decomposition using DECT for metabolic BM imaging. With BFC, the SUV, K1, and Ki values significantly increased by an average of 13.28% in BM regions compared to those without BFC (all P<0.0001), indicating the impact of BFC for BM quantification. Parametric imaging with BFC further confirmed regional analysis. Our study using DECT suggests current SUV and kinetic quantification of BM are likely underestimated in PET due to the presence of a significant bone volume fraction. Incorporating tissue composition information through BFC may improve BM metabolic quantification.

physics.med-ph↗

arXiv2025

Terahertz Signal Coverage Enhancement in Hall Scenarios Based on Single-Hop and Dual-Hop Reconfigurable Intelligent Surfaces

Terahertz (THz) communication offers ultra-high data rates and has emerged as a promising technology for future wireless networks. However, the inherently high free-space path loss of THz waves significantly limits the coverage range of THz communication systems. Therefore, extending the effective coverage area is a key challenge for the practical deployment of THz networks. Reconfigurable intelligent surfaces (RIS), which can dynamically manipulate electromagnetic wave propagation, provide a solution to enhance THz coverage. To investigate multi-RIS deployment scenarios, this work integrates an antenna array-based RIS model into the ray-tracing simulation platform. Using an indoor hall as a representative case study, the enhancement effects of single-hop and dual-hop RIS configurations on indoor signal coverage are evaluated under various deployment schemes. The developed framework offers valuable insights and design references for optimizing RIS-assisted indoor THz communication and coverage estimation.

eess.SP↗

arXiv2025

A User-customized and Untethered Electro-haptic Device for Immersive Human-Machine Interaction

Haptic feedback is essential for human-machine interaction, as it bridges physical and digital experiences and enables immersive engagement with virtual environments. However, current haptic devices are frequently tethered, lack portability and flexibility. They also have limited ability to deliver fine-grained, multi-dimensional feedback. To address these challenges, we present a flexible, ultra-thin, and user-customized electro-haptic device fabricated with soft materials and printable liquid metal ink. Its highly integrated and lightweight design minimizes interference with natural hand movements while maintaining reliable skin contact. By delivering finely controlled electrical stimulation through 15 electrodes, it can evoke a wide range of tactile sensations that cover diverse interaction scenarios. Our user study demonstrates that the device is comfortable to wear and capable of generating tunable, precise electro-haptic feedback, thereby significantly enhancing immersion and realism in human-machine interactions.

cs.HC↗

arXiv2025

Medical Referring Image Segmentation via Next-Token Mask Prediction

Medical Referring Image Segmentation (MRIS) involves segmenting target regions in medical images based on natural language descriptions. While achieving promising results, recent approaches usually involve complex design of multimodal fusion or multi-stage decoders. In this work, we propose NTP-MRISeg, a novel framework that reformulates MRIS as an autoregressive next-token prediction task over a unified multimodal sequence of tokenized image, text, and mask representations. This formulation streamlines model design by eliminating the need for modality-specific fusion and external segmentation models, supports a unified architecture for end-to-end training. It also enables the use of pretrained tokenizers from emerging large-scale multimodal models, enhancing generalization and adaptability. More importantly, to address challenges under this formulation-such as exposure bias, long-tail token distributions, and fine-grained lesion edges-we propose three novel strategies: (1) a Next-k Token Prediction (NkTP) scheme to reduce cumulative prediction errors, (2) Token-level Contrastive Learning (TCL) to enhance boundary sensitivity and mitigate long-tail distribution effects, and (3) a memory-based Hard Error Token (HET) optimization strategy that emphasizes difficult tokens during training. Extensive experiments on the QaTa-COV19 and MosMedData+ datasets demonstrate that NTP-MRISeg achieves new state-of-the-art performance, offering a streamlined and effective alternative to traditional MRIS pipelines.

cs.CV↗

arXiv2025

On finding gravitational waves from anisotropies of the Cosmic Microwave Background

The integrated Sachs-Wolfe (ISW) effect describes how photons are gravitationally redshifted, producing anisotropies in the Cosmic Microwave Background. We study the inverse problem and show that primordial gravitational perturbations, in particular their polarizations in the transversally traceless (TT) gauge can be identified from the local observation of the ISW effect.

gr-qc↗

arXiv2025

Gaussian process analysis of type-B quasiperiodic oscillations in the black hole X-ray binary MAXI J1348-630

We analyzed Insight-HXMT data of the black hole X-ray binary MAXI J1348-630 during the type-B QPO phase of its 2019 outburst. Using the Gaussian process method, we applied an additive composite kernel model consisting of an SHO, a DRW, and an additional white noise (AWN) to data from three energy bands: LE (1-10 keV), ME (10-30 keV), and HE (30-150 keV). We find that for the DRW component, correlations on the timescale of $τ_{\rm DRW}\sim10$ s are absent in the LE band, while they persist in the ME and HE bands over the full duration of the light curves. This energy-dependent behavior may reflect thermal instabilities, with the shorter correlation timescale in the disk compared to the corona. Alternatively, it may reflect variable Comptonizations of seed photons from different disk regions. Inner-disk photons are scattered by a small inner corona, producing soft X-rays. Outer-disk photons interact with an extended, jet-like corona, resulting in harder emission. The QPO is captured by an SHO component with a stable period of $\sim 0.2$ s and a high quality factor of $\sim 10$. The absence of significant evolution with energy or time of the SHO component suggests a connection between the accretion disk and the corona, which may be built by coherent oscillations of disk-corona driven by magnetorotational instability. The AWN components are present in all the three-band data and dominate over the DRW and SHO components. We interpret the AWN as another fast DRW with its $τ_{\rm DRW} < 0.01$ s. It may trace high-frequency fluctuations that occur in both the inner region of the accretion disk and the corona. Overall, our work reveals a timescale hierarchy in the coupled disk-corona scenario: fast DRW < SHO < disk DRW < corona DRW.

astro-ph.HE↗

arXiv2025

Rapid Scaling of Compositional Uncertainty from Sample to Population Levels

Understanding population composition is essential across ecological, evolutionary, conservation, and resource management contexts. Modern methods such as genetic stock identification (GSI) estimate the proportion of individuals from each subpopulation using genetic data. Ideally, these estimates are obtained through mixture analysis, which captures both sampling and genetic uncertainty. However, historical datasets often rely on individual assignment methods that only account for sample-level uncertainty, limiting the validity of population-level inferences. To address this, we propose a reverse Dirichlet-multinomial model and derive multiple variance estimators to propagate uncertainty from the sample to the population level. We extend this framework to genetic mark-recapture studies, assess performance via simulation, and apply our method to estimate the escapement of Sockeye Salmon (Oncorhynchus nerka) in the Taku River.

stat.ME↗

arXiv2025

Integrating Object Interaction Self-Attention and GAN-Based Debiasing for Visual Question Answering

Visual Question Answering (VQA) presents a unique challenge by requiring models to understand and reason about visual content to answer questions accurately. Existing VQA models often struggle with biases introduced by the training data, leading to over-reliance on superficial patterns and inadequate generalization to diverse questions and images. This paper presents a novel model, IOG-VQA, which integrates Object Interaction Self-Attention and GAN-Based Debiasing to enhance VQA model performance. The self-attention mechanism allows our model to capture complex interactions between objects within an image, providing a more comprehensive understanding of the visual context. Meanwhile, the GAN-based debiasing framework generates unbiased data distributions, helping the model to learn more robust and generalizable features. By leveraging these two components, IOG-VQA effectively combines visual and textual information to address the inherent biases in VQA datasets. Extensive experiments on the VQA-CP v1 and VQA-CP v2 datasets demonstrate that our model shows excellent performance compared with the existing methods, particularly in handling biased and imbalanced data distributions highlighting the importance of addressing both object interactions and dataset biases in advancing VQA tasks. Our code is available at https://github.com/HubuKG/IOG-VQA.

cs.CV↗

arXiv2025

MAJORScore: A Novel Metric for Evaluating Multimodal Relevance via Joint Representation

The multimodal relevance metric is usually borrowed from the embedding ability of pretrained contrastive learning models for bimodal data, which is used to evaluate the correlation between cross-modal data (e.g., CLIP). However, the commonly used evaluation metrics are only suitable for the associated analysis between two modalities, which greatly limits the evaluation of multimodal similarity. Herein, we propose MAJORScore, a brand-new evaluation metric for the relevance of multiple modalities ($N$ modalities, $N\ge3$) via multimodal joint representation for the first time. The ability of multimodal joint representation to integrate multiple modalities into the same latent space can accurately represent different modalities at one scale, providing support for fair relevance scoring. Extensive experiments have shown that MAJORScore increases by 26.03%-64.29% for consistent modality and decreases by 13.28%-20.54% for inconsistence compared to existing methods. MAJORScore serves as a more reliable metric for evaluating similarity on large-scale multimodal datasets and multimodal model performance evaluation.

cs.CV↗

arXiv2025

StereoAdapter: Adapting Stereo Depth Estimation to Underwater Scenes

Underwater stereo depth estimation provides accurate 3D geometry for robotics tasks such as navigation, inspection, and mapping, offering metric depth from low-cost passive cameras while avoiding the scale ambiguity of monocular methods. However, existing approaches face two critical challenges: (i) parameter-efficiently adapting large vision foundation encoders to the underwater domain without extensive labeled data, and (ii) tightly fusing globally coherent but scale-ambiguous monocular priors with locally metric yet photometrically fragile stereo correspondences. To address these challenges, we propose StereoAdapter, a parameter-efficient self-supervised framework that integrates a LoRA-adapted monocular foundation encoder with a recurrent stereo refinement module. We further introduce dynamic LoRA adaptation for efficient rank selection and pre-training on the synthetic UW-StereoDepth-40K dataset to enhance robustness under diverse underwater conditions. Comprehensive evaluations on both simulated and real-world benchmarks show improvements of 6.11% on TartanAir and 5.12% on SQUID compared to state-of-the-art methods, while real-world deployment with the BlueROV2 robot further demonstrates the consistent robustness of our approach. Code: https://github.com/AIGeeksGroup/StereoAdapter. Website: https://aigeeksgroup.github.io/StereoAdapter.

cs.CV↗

arXiv2025

A Tutorial on Conducting Mediation Analysis with Exposure Mixtures

Causal mediation analysis is a powerful tool in environmental health research, allowing researchers to uncover the pathways through which exposures influence health outcomes. While traditional mediation methods have been widely applied to individual exposures, real-world scenarios often involve complex mixtures. Such mixtures introduce unique methodological challenges, including multicollinearity, sparsity of active exposures, and potential nonlinear and interactive effects. This paper provides an overview of several commonly used approaches for mediation analysis under exposure mixture settings with clear strategies and code for implementation. The methods include: single exposure mediation analysis (SE-MA), principal component-based mediation analysis, environmental risk score-based mediation analysis, and Bayesian kernel machine regression causal mediation analysis. While SE-MA serves as a baseline that analyzes each exposure individually, the other methods are designed to address the correlation and complexity inherent in exposure mixtures. For each method, we aim to clarify the target estimand and the assumptions that each method is making to render a causal interpretation of the estimates obtained. We conduct a simulation study to systematically evaluate the operating characteristics of these four methods to estimate global indirect effects and to identify individual exposures contributing to the global mediation under varying sample sizes, effect sizes, and exposure-mediator-outcome structures. We also illustrate their real-world applicability by examining data from the PROTECT birth cohort, specifically analyzing the relationship between prenatal exposure to phthalate mixtures and neonatal head circumference Z-score, with leukotriene E4 as a mediator. This example offers practical guidance for conducting mediation analysis in complex environmental contexts.

stat.ME↗

arXiv2025

Towards Enhancing Data Equity in Public Health Data Science

Data-driven decisions shape public health policies and practice, yet persistent disparities in data representation skew insights and undermine interventions. To address this, we advance a structured roadmap that integrates public health data science with computer science and is grounded in reflexivity. We adopt data equity as a guiding concept: ensuring the fair and inclusive representation, collection, and use of data to prevent the introduction or exacerbation of systemic biases that could lead to invalid downstream inference and decisions. To underscore urgency, we present three public health cases where non-representative datasets and skewed knowledge impede decisions across diverse subgroups. These challenges echo themes in two literatures: public health highlights gaps in high-quality data for specific populations, while computer science and statistics contribute criteria and metrics for diagnosing bias in data and models. Building on these foundations, we propose a working definition of public health data equity and a structured self-audit framework. Our framework integrates core computational principles (fairness, accountability, transparency, ethics, privacy, confidentiality) with key public health considerations (selection bias, representativeness, generalizability, causality, information bias) to guide equitable practice across the data life cycle, from study design and data collection to measurement, analysis, interpretation, and translation. Embedding data equity in routine practice offers a practical path for ensuring that data-driven policies, artificial intelligence, and emerging technologies improve health outcomes for all. Finally, we emphasize the critical understanding that, although data equity is an essential first step, it does not inherently guarantee information, learning, or decision equity.

stat.AP↗