Author name cluster

Yuqing Wang

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

14 papers

1 author row

EAAI Journal 2026 Journal Article

An optimized hybrid deep learning approach with uncertainty quantification for accurate transformer winding hotspot temperature forecasting

Ali Abdo
Hongshun Liu
Yuqing Wang
Jiali Liu
Fuqiang Ren
Qingquan Li
Redhwan Algabri

Accurate forecasting of Winding Hotspot Temperature (WhotsptTem) is essential for reliable power transformer operation; however, complex non-linear behavior under varying load and environmental conditions pose significant modeling challenges. Conventional methods often yield limited accuracy in capturing the high thermal inertia and conditional interactions inherent in these assets. Furthermore, existing Deep Learning (DL) approaches are often hindered by sub-optimal manual hyperparameter tuning and typically produce only deterministic point forecasts, failing to quantify the reliability risks associated with prediction uncertainty. To address these limitations, this study proposes a novel Artificial Intelligence (AI) framework that employs a Genetic Algorithm (GA) to autonomously optimize a hybrid Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) architecture. This optimization strategy significantly enhances the model's ability to capture both high-frequency thermal transients and long-term dependencies, a necessity validated by dynamic feature analysis showing conditional load-ambient temperature interactions. Additionally, to provide statistically reliable probabilistic forecasts, a residual-based Monte Carlo (R-MC) simulation is integrated into the recursive multi-step forecasting process. The proposed Uncertainty Quantification (UQ) framework was validated by achieving a Prediction Interval Coverage Probability (PICP) of 93. 20% and a Mean Prediction Interval Width (MPIW) of 0. 3236 °C (°C). Extensive experiments on unseen test data demonstrate that the GA-optimized model outperforms state-of-the-art baselines, achieving a Root Mean Square Error (RMSE) of 0. 0908 °C and a Mean Absolute Percentage Error (MAPE) of 0. 24%. This research advances predictive maintenance strategies by providing high-precision, context-aware, and risk-quantified thermal forecasts.

Details DOI

JMLR Journal 2025 Journal Article

Good regularity creates large learning rate implicit biases: edge of stability, balancing, and catapult

Yuqing Wang
Zhenghao Xu
Tuo Zhao
Molei Tao

Large learning rates, when applied to gradient descent for nonconvex optimization, yield various implicit biases including the edge of stability, balancing, and catapult. These phenomena cannot be well explained by classical optimization theory. Though significant theoretical progress has been made in understanding these implicit biases, it remains unclear for which objective functions they are more likely to occur --- more precisely, for which functions there exists a larger set of initial conditions that lead to these phenomena? This paper provides an initial step in answering this question and also shows that these implicit biases are in fact various tips of the same iceberg. To establish these results, we develop a global convergence theory under large learning rates, for a family of nonconvex functions without globally Lipschitz continuous gradient, which was typically assumed in existing convergence analysis. Specifically, these phenomena are more likely to occur when the optimization objective function has good regularity. This regularity, together with gradient descent using a large learning rate that favors flatter regions, results in these nontrivial dynamical behaviors. Another corollary is the first non-asymptotic convergence rate bound for large-learning-rate gradient descent optimization of nonconvex functions. Although our theory only applies to specific functions so far, the possibility of extrapolating it to neural networks is also experimentally validated, for which different choices of loss, activation functions, and other techniques such as batch normalization can all affect regularity significantly and lead to very different training dynamics. [abs] [ pdf ][ bib ] &copy JMLR 2025. ( edit, beta )

PDF Details

NeurIPS Conference 2025 Conference Paper

Learning Generalizable Shape Completion with SIM(3) Equivariance

Yuqing Wang
Zhaiyu Chen
Xiaoxiang Zhu

3D shape completion methods typically assume scans are pre-aligned to a canonical frame. This leaks pose and scale cues that networks may exploit to memorize absolute positions rather than inferring intrinsic geometry. When such alignment is absent in real data, performance collapses. We argue that robust generalization demands architectural equivariance to the similarity group, SIM(3), so the model remains agnostic to pose and scale. Following this principle, we introduce the first SIM(3)-equivariant shape completion network, whose modular layers successively canonicalize features, reason over similarity-invariant geometry, and restore the original frame. Under a de-biased evaluation protocol that removes the hidden cues, our model outperforms both equivariant and augmentation baselines on the PCN benchmark. It also sets new cross-domain records on real driving and indoor scans, lowering minimal matching distance on KITTI by 17\% and Chamfer distance $\ell1$ on OmniObject3D by 14\%. Perhaps surprisingly, ours under the stricter protocol still outperforms competitors under their biased settings. These results establish full SIM(3) equivariance as an effective route to truly generalizable shape completion.

PDF Details

EAAI Journal 2025 Journal Article

Online prediction of hydro-pneumatic tensioner system of floating platform under internal waves

Jianwei Wang
Xiaofan Jin
Xuchu Liu
Ze He
Jiachen Chai
Pengfa Liu
Yuqing Wang
Wei Cai

To address the issue of low accuracy in the current motion response prediction model of the floating platform tensioner system, this paper proposes an online prediction method that integrates Empirical Mode Decomposition (EMD), Kernel Principal Component Analysis (KPCA), and Long Short-Term Memory (LSTM). The EMD technique is employed to decompose the sequence of environmental factors, reducing their non-stationarity. Subsequently, KPCA is used to extract key influencing factors and reduce input dimensionality. Finally, LSTM neural networks are applied to capture long-term dependencies in features and make accurate predictions. By validating the model using motion response data from the tensioner platform device under two scenarios with and without internal waves, it is compared against other models. The results show that the EMD-KPCA-LSTM model has high prediction accuracy in both scenarios. In particular, compared with the Convolutional Neural Network (CNN) model, the mean Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE) of the displacement and tension of the system decreased by 52. 97%, 55. 14%, 56. 31%, 68. 97%, 71. 02% 57. 60%, respectively, and R-square (R2) increased by 7. 14% and 12. 37%. In summary, the model has a good ability for data fitting and high prediction accuracy and has important practical value.

Details DOI

NeurIPS Conference 2025 Conference Paper

Understand Before You Generate: Self-Guided Training for Autoregressive Image Generation

Xiaoyu Yue
Zidong Wang
Yuqing Wang
Wenlong Zhang
Xihui Liu
Wanli Ouyang
Lei Bai
Luping Zhou

Recent studies have demonstrated the importance of high-quality visual representations in image generation and have highlighted the limitations of generative models in image understanding. As a generative paradigm originally designed for natural language, autoregressive models face similar challenges. In this work, we present the first systematic investigation into the mechanisms of applying the next-token prediction paradigm to the visual domain. We identify three key properties that hinder the learning of high-level visual semantics: local and conditional dependence, inter-step semantic inconsistency, and spatial invariance deficiency. We show that these issues can be effectively addressed by introducing self-supervised objectives during training, leading to a novel training framework, Self-guided Training for AutoRegressive models (ST-AR). Without relying on pre-trained representation models, ST-AR significantly enhances the image understanding ability of autoregressive models and leads to improved generation quality. Specifically, ST-AR brings approximately 42% FID improvement for LlamaGen-L and 49% FID improvement for LlamaGen-XL, while maintaining the same sampling strategy.

PDF Details

NeurIPS Conference 2024 Conference Paper

Evaluating the design space of diffusion-based generative models

Yuqing Wang
Ye He
Molei Tao

Most existing theoretical investigations of the accuracy of diffusion models, albeit significant, assume the score function has been approximated to a certain accuracy, and then use this a priori bound to control the error of generation. This article instead provides a first quantitative understanding of the whole generation process, i. e. , both training and sampling. More precisely, it conducts a non-asymptotic convergence analysis of denoising score matching under gradient descent. In addition, a refined sampling error analysis for variance exploding models is also provided. The combination of these two results yields a full error analysis, which elucidates (again, but this time theoretically) how to design the training and sampling processes for effective generation. For instance, our theory implies a preference toward noise distribution and loss weighting in training that qualitatively agree with the ones used in [Karras et al. , 2022]. It also provides perspectives on the choices of time and variance schedules in sampling: when the score is well trained, the design in [Song et al. , 2021] is more preferable, but when it is less trained, the design in [Karras et al. , 2022] becomes more preferable.

PDF Details DOI

NeurIPS Conference 2024 Conference Paper

LVD-2M: A Long-take Video Dataset with Temporally Dense Captions

Tianwei Xiong
Yuqing Wang
Daquan Zhou
Zhijie Lin
Jiashi Feng
Xihui Liu

The efficacy of video generation models heavily depends on the quality of their training datasets. Most previous video generation models are trained on short video clips, while recently there has been increasing interest in training long video generation models directly on longer videos. However, the lack of such high-quality long videos impedes the advancement long video generation. To promote research in long video generation, we desire a new dataset with four key features essential for training long video generation models: (1) long videos covering at least 10 seconds, (2) long-take videos without cuts, (3) large motion and diverse contents, and (4) temporally dense captions. To achieve this, we introduce a new pipeline for filtering high-quality long-take videos and generating temporally dense captions. Specifically, we define a set of metrics to quantitatively assess video quality including scene cuts, dynamic degrees, and semantic-level scores, enabling us to filter high-quality long-take videos from a large amount of source videos. Subsequently, we develop a hierarchical video captioning pipeline to annotate long videos with temporally-dense captions. With this pipeline, we curate the first long-take video dataset, LVD-2M, comprising 2 million long-take videos, each covering more than 10 seconds and annotated with temporally dense captions. We further validate the effectiveness of LVD-2M by fine-tuning video generation models to generate long videos with dynamic motions. We believe it will significantly contribute to future research in long video generation.

PDF Details DOI

NeurIPS Conference 2024 Conference Paper

Provable Acceleration of Nesterov's Accelerated Gradient for Asymmetric Matrix Factorization and Linear Neural Networks

Zhenghao Xu
Yuqing Wang
Tuo Zhao
Rachel Ward
Molei Tao

We study the convergence rate of first-order methods for rectangular matrix factorization, which is a canonical nonconvex optimization problem. Specifically, given a rank-$r$ matrix $\mathbf{A}\in\mathbb{R}^{m\times n}$, we prove that gradient descent (GD) can find a pair of $\epsilon$-optimal solutions $\mathbf{X}_T\in\mathbb{R}^{m\times d}$ and $\mathbf{Y}_T\in\mathbb{R}^{n\times d}$, where $d\geq r$, satisfying $\lVert\mathbf{X}_T\mathbf{Y}_T^\top-\mathbf{A}\rVert_F\leq\epsilon\lVert\mathbf{A}\rVert_F$ in $T=O(\kappa^2\log\frac{1}{\epsilon})$ iterations with high probability, where $\kappa$ denotes the condition number of $\mathbf{A}$. Furthermore, we prove that Nesterov's accelerated gradient (NAG) attains an iteration complexity of $O(\kappa\log\frac{1}{\epsilon})$, which is the best-known bound of first-order methods for rectangular matrix factorization. Different from small balanced random initialization in the existing literature, we adopt an unbalanced initialization, where $\mathbf{X}_0$ is large and $\mathbf{Y}_0$ is $0$. Moreover, our initialization and analysis can be further extended to linear neural networks, where we prove that NAG can also attain an accelerated linear convergence rate. In particular, we only require the width of the network to be greater than or equal to the rank of the output label matrix. In contrast, previous results achieving the same rate require excessive widths that additionally depend on the condition number and the rank of the input data matrix.

PDF Details DOI

EAAI Journal 2023 Journal Article

An integrating spherical fuzzy AHP and axiomatic design approach and its application in human–machine interface design evaluation

Qinghua Liu
Jiadui Chen
Kai Yang
Dan Liu
Ling He
Qing Qin
Yuqing Wang

Human–machine interface (HMI) design evaluation is critical in interactive product development because it directly affects the cost of subsequent design and user experiments. The evaluation information of HMI design mainly depends on the subjective perception and preference of experts, especially the hesitancy degree is rarely considered. We propose an integrated spherical fuzzy AHP (SF-AHP) and spherical fuzzy axiomatic design (SF-AD) method to choose a reasonable HMI alternative, considering the potential risks caused by the hesitancy degree of experts. Firstly, we introduce a process of automatically repairing inconsistent spherical fuzzy preference relations (SFPR) for SF-AHP. After that, we build suitable evaluation criteria and calculate the criteria weight by the SF-AHP method. We further extend the axiomatic design to the spherical fuzzy environment and propose a multi-criteria decision-making (MCDM) method based on SF-AD to evaluate HMI alternatives. The case analysis results demonstrate the effectiveness of the proposed method in the HMI design evaluation process, and the sensitivity and comparative analysis results indicate that the proposed method is stable and reliable.

Details DOI

IJCAI Conference 2023 Conference Paper

Discovering Sounding Objects by Audio Queries for Audio Visual Segmentation

Shaofei Huang
Han Li
Yuqing Wang
Hongji Zhu
Jiao Dai
Jizhong Han
Wenge Rong
Si Liu

Audio visual segmentation (AVS) aims to segment the sounding objects for each frame of a given video. To distinguish the sounding objects from silent ones, both audio-visual semantic correspondence and temporal interaction are required. The previous method applies multi-frame cross-modal attention to conduct pixel-level interactions between audio features and visual features of multiple frames simultaneously, which is both redundant and implicit. In this paper, we propose an Audio-Queried Transformer architecture, AQFormer, where we define a set of object queries conditioned on audio information and associate each of them to particular sounding objects. Explicit object-level semantic correspondence between audio and visual modalities is established by gathering object information from visual features with predefined audio queries. Besides, an Audio-Bridged Temporal Interaction module is proposed to exchange sounding object-relevant information among multiple frames with the bridge of audio features. Extensive experiments are conducted on two AVS benchmarks to show that our method achieves state-of-the-art performances, especially 7. 1% M_J and 7. 6% M_F gains on the MS3 setting.

PDF Details DOI

NeurIPS Conference 2021 Conference Paper

Twins: Revisiting the Design of Spatial Attention in Vision Transformers

Xiangxiang Chu
Zhi Tian
Yuqing Wang
Bo Zhang
Haibing Ren
Xiaolin Wei
Huaxia Xia
Chunhua Shen

Very recently, a variety of vision transformer architectures for dense prediction tasks have been proposed and they show that the design of spatial attention is critical to their success in these tasks. In this work, we revisit the design of the spatial attention and demonstrate that a carefully devised yet simple spatial attention mechanism performs favorably against the state-of-the-art schemes. As a result, we propose two vision transformer architectures, namely, Twins- PCPVT and Twins-SVT. Our proposed architectures are highly efficient and easy to implement, only involving matrix multiplications that are highly optimized in modern deep learning frameworks. More importantly, the proposed architectures achieve excellent performance on a wide range of visual tasks including image-level classification as well as dense detection and segmentation. The simplicity and strong performance suggest that our proposed architectures may serve as stronger backbones for many vision tasks.

PDF Details

NeurIPS Conference 2020 Conference Paper

Why Do Deep Residual Networks Generalize Better than Deep Feedforward Networks? --- A Neural Tangent Kernel Perspective

Kaixuan Huang
Yuqing Wang
Molei Tao
Tuo Zhao

Deep residual networks (ResNets) have demonstrated better generalization performance than deep feedforward networks (FFNets). However, the theory behind such a phenomenon is still largely unknown. This paper studies this fundamental problem in deep learning from a so-called ``neural tangent kernel'' perspective. Specifically, we first show that under proper conditions, as the width goes to infinity, training deep ResNets can be viewed as learning reproducing kernel functions with some kernel function. We then compare the kernel of deep ResNets with that of deep FFNets and discover that the class of functions induced by the kernel of FFNets is asymptotically not learnable, as the depth goes to infinity. In contrast, the class of functions induced by the kernel of ResNets does not exhibit such degeneracy. Our discovery partially justifies the advantages of deep ResNets over deep FFNets in generalization abilities. Numerical results are provided to support our claim.

PDF Details

AAMAS Conference 2018 Conference Paper

A Geometric Least Squares Method for Peer Assessment

Mingyu Xiao
Yuqing Wang
Binglin Tao

In the peer assessment problem, a set of agents give evaluations to each other, and we are going to combine these peer assessments together to construct an overall evaluation. In this paper, we propose a geometric least squares method (GLS) to ﬁnd an aggregate scoring overall agents for the peer assessment problem. Our method is based on the following observation. Since each agent has a missing score that should be given by itself, we consider the missing score as a variable and then each agent can be regarded as a line in an n-dimensional vector space. The ﬁnal aggregate scores of the agents can be regarded as points on a line vector, called the projection vector. Thus, we treat the peer assessment problem as an optimization problem of selecting a projection vector with minimum total squared distance to all the lines representing the agents. We will see that this aggregate method has some advantages compared with the simple average method. One advantage is that, when the scores given by each agent (even ignoring the magnitude of the agent) are close to a groundtruth, the new method ﬁnds the groundtruth with the highest expectation.

PDF

IJCAI Conference 2017 Conference Paper

Score Aggregation via Spectral Method

Mingyu Xiao
Yuqing Wang

The score aggregation problem is to find an aggregate scoring over all candidates given individual scores provided by different agents. This is a fundamental problem with a broad range of applications in social choice and many other areas. The simple and commonly used method is to sum up all scores of each candidate, which is called the sum-up method. In this paper, we give good algebraic and geometric explanations for score aggregation, and develop a spectral method for it. If we view the original scores as `noise data', our method can find an `optimal' aggregate scoring by minimizing the `noise information'. We also suggest a signal-to-noise indicator to evaluate the validity of the aggregation or the consistency of the agents.

PDF Details