arXiv subjects

Qingyuan Yang

Publications and source records attributed to Qingyuan Yang.

16 recordsLinked to original sources

arXiv2026

FitAQA: A Benchmark of Fitness Action Quality Assessment for Multimodal Large Language Models

Fitness Action Quality Assessment (AQA) is important for intelligent sports training, yet the capabilities of Multimodal Large Language Models (MLLMs) in this setting remain underexplored. Existing benchmarks rely on action-specific annotation schemes and focus primarily on final assessment outputs, offering limited insight into how models assess exercise quality. We introduce FitAQA, a systematic benchmark for evaluating MLLMs in fitness AQA, containing 2,219 videos and 5,512 QA instances across 30 bodyweight exercises. In collaboration with experts in sports science, we develop a unified form error taxonomy that defines 38 recurring form errors within six complementary quality dimensions: alignment, symmetry, stability, coordination, tempo, and completeness. This taxonomy provides a shared assessment framework across different exercises. FitAQA further formulates three evaluation tasks: perception for recognizing relevant visual evidence, judgement for combining that evidence with domain knowledge to assess execution correctness, and temporal grounding for localizing form errors over time. Extensive evaluation shows that current MLLMs still struggle to assess exercise quality comprehensively and localize form errors precisely. Controlled experiments further indicate that visual perception is a key bottleneck, as judgement performance improves substantially when ground-truth perceptual evidence is provided. The dataset and evaluation code will be made publicly available.

cs.AI↗

arXiv2026

FRWKV+: Periodic-Aware Adaptive Gating for Frequency-Space Linear Time Series Forecasting

Accurate and efficient long-term multivariate time series forecasting requires capturing recurring temporal structure while keeping inference cheap across many variables and horizons. Frequency-space models represent long-range and periodic variation compactly, but they typically process the real and imaginary spectral components as weakly coupled streams and treat periodic cues as ordinary input features, even when such cues are unreliable. This paper proposes FRWKV-Plus, a lightweight periodic-aware frequency-space forecasting model built on the efficient FRWKV backbone. FRWKV-Plus introduces a cross-branch spectral gate that reweights each spectral branch using a summary of its sibling branch, and a trust-gated residual correction that converts compact within-period context into a bounded, sign-flexible adjustment of these gates under a learned, data-dependent trust score. By construction, the correction is identity-preserving at initialization and strictly bounded, so periodic evidence can refine but never dominate or invert the base interaction. On seven standard benchmarks, FRWKV-Plus is consistently competitive with strong linear, frequency-domain, recurrent-style, and Transformer-based forecasters while preserving the lightweight profile of the backbone. Controlled three-seed ablations show that each component contributes, that the benefit is modest on strongly periodic data and pronounced on the harder Exchange and ILI datasets, and that the within-period context is the most influential single component. The implementation is publicly available at https://github.com/yangqingyuan-byte/FRWKV-plus.

cs.LG↗

arXiv2025

FRWKV:Frequency-Domain Linear Attention for Long-Term Time Series Forecasting

Traditional Transformers face a major bottleneck in long-sequence time series forecasting due to their quadratic complexity $(\mathcal{O}(T^2))$ and their limited ability to effectively exploit frequency-domain information. Inspired by RWKV's $\mathcal{O}(T)$ linear attention and frequency-domain modeling, we propose FRWKV, a frequency-domain linear-attention framework that overcomes these limitations. Our model integrates linear attention mechanisms with frequency-domain analysis, achieving $\mathcal{O}(T)$ computational complexity in the attention path while exploiting spectral information to enhance temporal feature representations for scalable long-sequence modeling. Across eight real-world datasets, FRWKV achieves a first-place average rank. Our ablation studies confirm the critical roles of both the linear attention and frequency-encoder components. This work demonstrates the powerful synergy between linear attention and frequency analysis, establishing a new paradigm for scalable time series modeling. Code is available at this repository: https://github.com/yangqingyuan-byte/FRWKV.

cs.LG↗

arXiv2025

Every Activation Boosted: Scaling General Reasoner to 1 Trillion Open Language Foundation

We introduce Ling 2.0, a series reasoning-oriented language foundation built upon the principle that every activation boosts reasoning capability. Designed to scale from tens of billions to one trillion parameters under a unified Mixture-of-Experts (MoE) paradigm, Ling 2.0 emphasizes high sparsity, cross-scale consistency, and efficiency guided by empirical scaling laws. The series includes three non-thinking (instruct) models - Ling-mini-2.0, Ling-flash-2.0, and Ling-1T - ranging from 16B to 1T total parameters and achieving up to 7-fold active-compute efficiency compared with dense counterparts. Ling 2.0 integrates coordinated innovations across model architecture, pre-training, post-training, and infrastructure: a high-sparsity MoE with MTP for efficient reasoning, reasoning-oriented data and mid-training CoT activation, reinforcement-based fine-tuning (DFT, Evo-CoT), and full-scale FP8 training with fine-grained heterogeneous pipelines. At the trillion scale, Ling-1T establishes a new Pareto frontier of reasoning accuracy versus computational efficiency, demonstrating that sparse activation, when properly aligned with reasoning objectives, enables scalable and efficient intelligence. Collectively, Ling 2.0 provides a coherent, open, and efficient foundation for advancing future reasoning and thinking models, including the Ring series built upon the same base.

cs.CL↗

arXiv2025

Every Attention Matters: An Efficient Hybrid Architecture for Long-Context Reasoning

In this technical report, we present the Ring-linear model series, specifically including Ring-mini-linear-2.0 and Ring-flash-linear-2.0. Ring-mini-linear-2.0 comprises 16B parameters and 957M activations, while Ring-flash-linear-2.0 contains 104B parameters and 6.1B activations. Both models adopt a hybrid architecture that effectively integrates linear attention and softmax attention, significantly reducing I/O and computational overhead in long-context inference scenarios. Compared to a 32 billion parameter dense model, this series reduces inference cost to 1/10, and compared to the original Ring series, the cost is also reduced by over 50%. Furthermore, through systematic exploration of the ratio between different attention mechanisms in the hybrid architecture, we have identified the currently optimal model structure. Additionally, by leveraging our self-developed high-performance FP8 operator library-linghe, overall training efficiency has been improved by 50%. Benefiting from the high alignment between the training and inference engine operators, the models can undergo long-term, stable, and highly efficient optimization during the reinforcement learning phase, consistently maintaining SOTA performance across multiple challenging complex reasoning benchmarks.

cs.LG↗

arXiv2025

MSSDF: Modality-Shared Self-supervised Distillation for High-Resolution Multi-modal Remote Sensing Image Learning

Remote sensing image interpretation plays a critical role in environmental monitoring, urban planning, and disaster assessment. However, acquiring high-quality labeled data is often costly and time-consuming. To address this challenge, we proposes a multi-modal self-supervised learning framework that leverages high-resolution RGB images, multi-spectral data, and digital surface models (DSM) for pre-training. By designing an information-aware adaptive masking strategy, cross-modal masking mechanism, and multi-task self-supervised objectives, the framework effectively captures both the correlations across different modalities and the unique feature structures within each modality. We evaluated the proposed method on multiple downstream tasks, covering typical remote sensing applications such as scene classification, semantic segmentation, change detection, object detection, and depth estimation. Experiments are conducted on 15 remote sensing datasets, encompassing 26 tasks. The results demonstrate that the proposed method outperforms existing pretraining approaches in most tasks. Specifically, on the Potsdam and Vaihingen semantic segmentation tasks, our method achieved mIoU scores of 78.30\% and 76.50\%, with only 50\% train-set. For the US3D depth estimation task, the RMSE error is reduced to 0.182, and for the binary change detection task in SECOND dataset, our method achieved mIoU scores of 47.51\%, surpassing the second CS-MAE by 3 percentage points. Our pretrain code, checkpoints, and HR-Pairs dataset can be found in https://github.com/CVEO/MSSDF.

cs.CV↗

arXiv2024

Relax and Merge: A Simple Yet Effective Framework for Solving Fair $k$-Means and $k$-sparse Wasserstein Barycenter Problems

The fairness of clustering algorithms has gained widespread attention across various areas, including machine learning, In this paper, we study fair $k$-means clustering in Euclidean space. Given a dataset comprising several groups, the fairness constraint requires that each cluster should contain a proportion of points from each group within specified lower and upper bounds. Due to these fairness constraints, determining the optimal locations of $k$ centers is a quite challenging task. We propose a novel ``Relax and Merge'' framework that returns a $(1+4ρ+ O(ε))$-approximate solution, where $ρ$ is the approximate ratio of an off-the-shelf vanilla $k$-means algorithm and $O(ε)$ can be an arbitrarily small positive number. If equipped with a PTAS of $k$-means, our solution can achieve an approximation ratio of $(5+O(ε))$ with only a slight violation of the fairness constraints, which improves the current state-of-the-art approximation guarantee. Furthermore, using our framework, we can also obtain a $(1+4ρ+O(ε))$-approximate solution for the $k$-sparse Wasserstein Barycenter problem, which is a fundamental optimization problem in the field of optimal transport, and a $(2+6ρ)$-approximate solution for the strictly fair $k$-means clustering with no violation, both of which are better than the current state-of-the-art methods. In addition, the empirical results demonstrate that our proposed algorithm can significantly outperform baseline approaches in terms of clustering cost.

cs.LG↗

arXiv2024

A simple emulator that enables interpretation of parameter-output relationships, applied to two climate model PPEs

We present a new additive method, nicknamed sage for Simplified Additive Gaussian processes Emulator, to emulate climate model Perturbed Parameter Ensembles (PPEs). It estimates the value of a climate model output as the sum of additive terms. Each additive term is the mean of a Gaussian Process, and corresponds to the impact of a parameter or parameter group on the variable of interest. This design caters to the sparsity of PPEs which are characterized by limited ensemble members and high dimensionality of the parameter space. sage quantifies the variability explained by different parameters and parameter groups, providing additional insights on the parameter-climate model output relationship. We apply the method to two climate model PPEs and compare it to a fully connected Neural Network. The two methods have comparable performance with both PPEs, but sage provides insights on parameter and parameter group importance as well as diagnostics useful for optimizing PPE design. Insights gained are valid regardless of the emulator method used, and have not been previously addressed. Our work highlights that analyzing the PPE used to train an emulator is different from analyzing data generated from an emulator trained on the PPE, as the former provides more insights on the data structure in the PPE which could help inform the emulator design.

stat.ME↗

arXiv2024

S3Net: Innovating Stereo Matching and Semantic Segmentation with a Single-Branch Semantic Stereo Network in Satellite Epipolar Imagery

Stereo matching and semantic segmentation are significant tasks in binocular satellite 3D reconstruction. However, previous studies primarily view these as independent parallel tasks, lacking an integrated multitask learning framework. This work introduces a solution, the Single-branch Semantic Stereo Network (S3Net), which innovatively combines semantic segmentation and stereo matching using Self-Fuse and Mutual-Fuse modules. Unlike preceding methods that utilize semantic or disparity information independently, our method dentifies and leverages the intrinsic link between these two tasks, leading to a more accurate understanding of semantic information and disparity estimation. Comparative testing on the US3D dataset proves the effectiveness of our S3Net. Our model improves the mIoU in semantic segmentation from 61.38 to 67.39, and reduces the D1-Error and average endpoint error (EPE) in disparity estimation from 10.051 to 9.579 and 1.439 to 1.403 respectively, surpassing existing competitive methods. Our codes are available at:https://github.com/CVEO/S3Net.

cs.CV↗

arXiv2024

Approximate Algorithms For $k$-Sparse Wasserstein Barycenter With Outliers

Wasserstein Barycenter (WB) is one of the most fundamental optimization problems in optimal transportation. Given a set of distributions, the goal of WB is to find a new distribution that minimizes the average Wasserstein distance to them. The problem becomes even harder if we restrict the solution to be ``$k$-sparse''. In this paper, we study the $k$-sparse WB problem in the presence of outliers, which is a more practical setting since real-world data often contains noise. Existing WB algorithms cannot be directly extended to handle the case with outliers, and thus it is urgently needed to develop some novel ideas. First, we investigate the relation between $k$-sparse WB with outliers and the clustering (with outliers) problems. In particular, we propose a clustering based LP method that yields constant approximation factor for the $k$-sparse WB with outliers problem. Further, we utilize the coreset technique to achieve the $(1+ε)$-approximation factor for any $ε>0$, if the dimensionality is not high. Finally, we conduct the experiments for our proposed algorithms and illustrate their efficiencies in practice.

cs.LG↗

arXiv2023

On Robust Wasserstein Barycenter: The Model and Algorithm

The Wasserstein barycenter problem is to compute the average of $m$ given probability measures, which has been widely studied in many different areas; however, real-world data sets are often noisy and huge, which impedes its applications in practice. Hence, in this paper, we focus on improving the computational efficiency of two types of robust Wasserstein barycenter problem (RWB): fixed-support RWB (fixed-RWB) and free-support RWB (free-RWB); actually, the former is a subroutine of the latter. Firstly, we improve efficiency through model reducing; we reduce RWB as an augmented Wasserstein barycenter problem, which works for both fixed-RWB and free-RWB. Especially, fixed-RWB can be computed within $\widetilde{O}(\frac{mn^2}{ε_+})$ time by using an off-the-shelf solver, where $ε_+$ is the pre-specified additive error and $n$ is the size of locations of input measures. Then, for free-RWB, we leverage a quality guaranteed data compression technique, coreset, to accelerate computation by reducing the data set size $m$. It shows that running algorithms on the coreset is enough instead of on the original data set. Next, by combining the model reducing and coreset techniques above, we propose an algorithm for free-RWB by updating the weights and locations alternatively. Finally, our experiments demonstrate the efficiency of our techniques.

cs.LG↗

arXiv2022

Coresets for Relational Data and The Applications

A coreset is a small set that can approximately preserve the structure of the original input data set. Therefore we can run our algorithm on a coreset so as to reduce the total computational complexity. Conventional coreset techniques assume that the input data set is available to process explicitly. However, this assumption may not hold in real-world scenarios. In this paper, we consider the problem of coresets construction over relational data. Namely, the data is decoupled into several relational tables, and it could be very expensive to directly materialize the data matrix by joining the tables. We propose a novel approach called ``aggregation tree with pseudo-cube'' that can build a coreset from bottom to up. Moreover, our approach can neatly circumvent several troublesome issues of relational learning problems [Khamis et al., PODS 2019]. Under some mild assumptions, we show that our coreset approach can be applied for the machine learning tasks, such as clustering, logistic regression and SVM.

cs.LG↗

arXiv2021

The development of near-vent volcanic ash cloud layers due to inhomogeneous atmospheric turbulence and relationship to wind shear

physics.ao-ph↗

arXiv2020

The development of volcanic ash cloud layers over hours to days due to turbulence layering

Volcanic ash clouds often become multilayered and thin with distance from the vent. We explore one mechanism for development of this layered structure. We review data on the characteristics of turbulence layering in the free atmosphere, as well as examples of observations of layered clouds both near-vent and distally. We then explore dispersion models that explicitly use the observed layered structure of atmospheric turbulence. The results suggest that the alternation of turbulent and quiescent atmospheric layers provides one mechanism for development of multilayered ash clouds by modulating vertical particle motion. The largest particles, generally $> 100 μ$m, are little affected by turbulence. For particles in which both settling and turbulent diffusion are important to vertical motion, mostly in the range of 10-100 $μ$m, the greater turbulence intensity and more rapid turbulent diffusion in some layers causes these particles to spend greater time in the more turbulent layers, leading to a layering of concentration. For smaller particles, mostly in the submicron range, the more rapid diffusion in the turbulent layers causes these particles to ``wash out'' quickly.

physics.ao-ph↗

arXiv2019

Stratigraphic and sedimentologic framework for tephras in the Wilson Creek Formation, Mono Basin, California, USA

Numerous tephra layers occur within the late Pleistocene Wilson Creek Formation, where they are interbedded with lacustrine deposits of Lake Russell, the ancestor of present-day Mono Lake. Most of the tephra layers are rhyolitic in composition, and were produced from the Mono Craters. We present detailed stratigraphy and sedimentology of the tephra layers, sampled at twelve outcrops near the shoreline of Mono Lake and the Mono Craters, and implement grain size, componentry, and surface morphology analysis to characterize their physical properties. Sub-unit correlation is proposed for certain tephra units. Noticeable features of the tephras, such as the occurrence of low-density rounded or highly vesicular pumice within certain sub-units, are highlighted. The abundant obsidian, lithics, and ostracods within many sub-units suggest that the associated eruption pulses involved water-magma interaction. Eruptions from the Mono Craters during the late Pleistocene were consistent neither in frequency nor volume. The Mono Craters were most active during the eruption of tephras in Sequences C ($\sim$42.5-39.4 ka) and A ($\sim$14.1-12.9 ka), and Pleistocene volcanic activity reached its peak during the eruption of tephra layer C11. Detailed interpretation of tephra layer B7 and tephras in Sequence A is given. Tephras in Sequence B ($\sim$26.2-23.1 ka) may have had their vents located in the southern half of the Mono Craters, or are smaller in volume (except for B7), compared to the other tephras in the Wilson Creek Formation. Vents for tephra layers A4, A3, and A1 are located near the northern end of the Mono Craters. The stratigraphy of tephra layers B7 and A1 suggests an unstable depositional environment, which can be used to help constrain the water-level history of Lake Russell during the late Pleistocene.

physics.geo-ph↗

arXiv2018

A new method to identify the source vent location of tephra fall deposits: development and testing, and application to key Quaternary eruptions of Western North America

A new method to identify the source vent location of tephra fall deposits based on thickness or maximum clast size measurements is presented in this work. It couples a first-order gradient descent method with either one of two commonly-used semi-empirical models of tephra thickness distribution. The method is successfully applied to three tephra thickness and one maximum clast size datasets of the North Mono and Fogo A tephra deposits. Randomly selected and localized subsets of these datasets are used as input to evaluate its performance in cases of sparse observations. The results suggest that estimating the dispersal axis is a more robust way to constrain the source vent location with sparse observations. Bootstrap aggregating and examining the surface of the cost function are proposed to characterize the uncertainty of the method. Distinctions between the two adopted semi-empirical models of tephra thickness distribution are discussed. Results from applying the method to thickness datasets of the Trego Hot Springs and Rockland tephras are consistent with previous studies, which also provide new estimates on their total volume. The method is then applied to a series of well-correlated tephra sub-units preserved within the Wilson Creek Formation to estimate their vent location and total volume. The simplicity and flexibility of the method make it a potentially useful and powerful tool for the study of tephra fall deposits of different characteristics.

physics.geo-ph↗