Feng Xiao

AAAI Conference 2026 Conference Paper

CyC3D: Fine-grained Controllable 3D Generation via Cycle Consistency Regularization

• Hongbin Xu
• Chaohui Yu
• Feng Xiao
• Jiazheng Xing
• Hai Ci
• Weitao Chen
• Fan Wang
• Ming Li

Despite the remarkable progress of 3D generation, achieving controllability, i.e., ensuring consistency between generated 3D content and input conditions like edge and depth, remains a significant challenge. Existing methods often struggle to maintain accurate alignment, leading to noticeable discrepancies. To address this issue, we propose CyC3D, a new framework that enhances controllable 3D generation by explicitly encouraging cyclic consistency between the second-order 3D content, generated based on extracted signals from the first-order generation, and its original input controls. Specifically, we employ an efficient feed-forward backbone that can generate a 3D object from an input condition and a text prompt. Given an initial viewpoint and a control signal, a novel view is rendered from the generated 3D content, from which the extracted condition is used to regenerate the 3D content. This re-generated output is then rendered back to the initial viewpoint, followed by another round of control signal extraction, forming a cyclic process with two consistency constraints. View consistency ensures coherence between the two generated 3D objects, measured by semantic similarity to accommodate generative diversity. Condition consistency aligns the final extracted signal with the original input control, preserving structural or geometric details throughout the process. Extensive experiments on popular benchmarks demonstrate that CyC3D significantly improves controllability, especially for fine-grained details, outperforming existing methods across various conditions (e.g., +14.17% PSNR for edge, +6.26% PSNR for sketch).

EAAI Journal 2026 Journal Article

High-precision multimodal vehicle trajectory prediction model based on cross-layer interleaved spatiotemporal attention mechanism

• Fei Teng
• Liqiang Jin
• Junnian Wang
• Feng Xiao
• Mengdi Guo
• Yanbo Zhou
• Jin Zhang

In increasingly complex traffic environments, spatiotemporal attention mechanisms have made remarkable advancements in scene-level interaction modelling. However, the deep and multi-scale spatiotemporal representations required for safe and efficient decision-making in intelligent vehicles remain underexplored. Aiming to address this limitation, this study proposes a multimodal trajectory prediction model based on a cross-layer interleaved spatiotemporal attention (CLISTA) mechanism. Compared with conventional spatiotemporal attention frameworks, CLISTA more effectively captures multi-scale spatiotemporal interactions in complex traffic scenes through the alternating fusion of spatial and temporal features across network layers via a cross-layer interleaving structure. Firstly, spatial, dynamic and heading conflict risks are derived from the relative motion between the target vehicle and its neighbours and aggregated into a social grid weight matrix, through which the neighbours' collective influence on the target vehicle is quantified. Secondly, spatial and temporal multi-head attention modules are designed within each layer. By integrating an interleaved ‘spatial–temporal’ stacking strategy with cross-layer skip connections, the model facilitates progressive alignment and deep fusion, ranging from local interactions to long-range dependencies. Subsequently, an intention recognition module is developed. A second-order gated bilinear fusion mechanism is introduced to adaptively model higher-order couplings between local neighbour dynamics and global interaction semantics, thereby yielding a multimodal probability distribution over the target vehicle's driving intentions. Lastly, multimodal trajectory predictions are generated by decoding the fused spatiotemporal features together with the inferred intention information. Experimental results on three benchmark datasets—NGSIM (Next Generation Simulation), AD4CHE (Aerial Dataset for China Congested Highway and Expressway), and highD—demonstrate that CLISTA consistently outperforms the baseline methods. Relative to the next-best model, it reduces average/final displacement errors by 16. 67 %/21. 23 %, 12. 99 %/21. 14 % and 10. 53 %/21. 59 % on NGSIM, AD4CHE and HighD, respectively. Overall, CLISTA offers reliable multi-hypothesis trajectory priors for safe and efficient decision-making in complex traffic scenarios.

EAAI Journal 2025 Journal Article

An expert features enhanced temporal and contextual contrasting learning model for detecting wind turbine blade icing

• Jiamei Zhou
• Feng Xiao
• Xiaoying Zhang
• Xu Cheng
• Jianhua Zhang

With global carbon neutrality goals, wind power has rapidly developed, but blade icing remains a major challenge. AI(artificial intelligence) methods show great promise for detecting icing on wind turbine blades. However, early icing data overlap, difficulty obtaining continuous labeled data, and small variations between samples due to short sampling intervals complicate the task. This study proposes an expert feature-enhanced temporal and contextual contrastive learning model for detecting blade icing. This approach efficiently extracts data features and combines self-supervised contrastive learning, maximizing data utilization without requiring extensive labeled data. To validate the effectiveness of this method, extensive experiments were conducted on two public datasets. The results achieved the best performance across multiple metrics, with F1-Score and AUC exceeding 98%, significantly enhancing wind power generation efficiency.

NeurIPS Conference 2025 Conference Paper

Domain-Specific Pruning of Large Mixture-of-Experts Models with Few-shot Demonstrations

• Zican Dong
• Han Peng
• Peiyu Liu
• Xin Zhao
• Dong Wu
• Feng Xiao
• Zhifeng Wang

Mixture-of-Experts (MoE) models achieve a favorable trade-off between performance and inference efficiency by activating only a subset of experts. However, the memory overhead of storing all experts remains a major limitation, especially in large-scale MoE models such as DeepSeek-R1 (671B). In this study, we investigate domain specialization and expert redundancy in large-scale MoE models and uncover a consistent behavior we term~\emph{few-shot expert localization}, with only a few in-domain demonstrations, the model consistently activates a sparse and stable subset of experts on tasks within the same domain. Building on this observation, we propose a simple yet effective pruning framework, \textbf{EASY-EP}, that leverages a few domain-specific demonstrations to identify and retain only the most relevant experts. EASY-EP comprises two key components: \textbf{output-aware expert importance assessment} and \textbf{expert-level token contribution estimation}. The former evaluates the importance of each expert for the current token by considering the gating scores and L2 norm of the outputs of activated experts, while the latter assesses the contribution of tokens based on representation similarities before and after routed experts. Experiments on DeepSeek-R1 and DeepSeek-V3-0324 show that our method can achieve comparable performances and $2. 99\times$ throughput under the same memory budget as the full model, with only half the experts. Our code is available at https: //github. com/RUCAIBox/EASYEP.

NeurIPS Conference 2025 Conference Paper

Fairness-aware Anomaly Detection via Fair Projection

• Feng Xiao
• Xiaoying Tang
• Jicong Fan

Unsupervised anomaly detection is a critical task in many high-social-impact applications such as finance, healthcare, social media, and cybersecurity, where demographics involving age, gender, race, disease, etc. are used frequently. In these scenarios, possible bias from anomaly detection systems can lead to unfair treatment for different groups and even exacerbate social bias. In this work, first, we thoroughly analyze the feasibility and necessary assumptions for ensuring group fairness in unsupervised anomaly detection. Second, we propose a novel fairness-aware anomaly detection method FairAD. From the normal training data, FairAD learns a projection to map data of different demographic groups to a common target distribution that is simple and compact, and hence provides a reliable base to estimate the density of the data. The density can be directly used to identify anomalies while the common target distribution ensures fairness between different groups. Furthermore, we propose a threshold-free fairness metric that provides a global view for model's fairness, eliminating dependence on manual threshold selection. Experiments on real-world benchmarks demonstrate that our method achieves an improved trade-off between detection accuracy and fairness under both balanced and skewed data across different groups.

NeurIPS Conference 2024 Conference Paper

Unsupervised Anomaly Detection in The Presence of Missing Values

• Feng Xiao
• Jicong Fan

Anomaly detection methods typically require fully observed data for model training and inference and cannot handle incomplete data, while the missing data problem is pervasive in science and engineering, leading to challenges in many important applications such as abnormal user detection in recommendation systems and novel or anomalous cell detection in bioinformatics, where the missing rates can be higher than 30\% or even 80\%. In this work, first, we construct and evaluate a straightforward strategy, ''impute-then-detect'', via combining state-of-the-art imputation methods with unsupervised anomaly detection methods, where the training data are composed of normal samples only. We observe that such two-stage methods frequently yield imputation bias from normal data, namely, the imputation methods are inclined to make incomplete samples ''normal", where the fundamental reason is that the imputation models learned only on normal data and cannot generalize well to abnormal data in the inference stage. To address this challenge, we propose an end-to-end method that integrates data imputation with anomaly detection into a unified optimization problem. The proposed model learns to generate well-designed pseudo-abnormal samples to mitigate the imputation bias and ensure the discrimination ability of both the imputation and detection processes. Furthermore, we provide theoretical guarantees for the effectiveness of the proposed method, proving that the proposed method can correctly detect anomalies with high probability. Experimental results on datasets with manually constructed missing values and inherent missing values demonstrate that our proposed method effectively mitigates the imputation bias and surpasses the baseline methods significantly. The source code of our method is available at https: //github. com/jicongfan/ImAD-Anomaly-Detection-With-Missing-Data.

YNIMG Journal 2023 Journal Article

Deep learning-assisted identification and quantification of aneurysmal subarachnoid hemorrhage in non-contrast CT scans: Development and external validation of Hybrid 2D/3D UNet

• Ping Hu
• Haizhu Zhou
• Tengfeng Yan
• Hongping Miu
• Feng Xiao
• Xinyi Zhu
• Lei Shu
• Shuang Yang

Accurate stroke assessment and consequent favorable clinical outcomes rely on the early identification and quantification of aneurysmal subarachnoid hemorrhage (aSAH) in non-contrast computed tomography (NCCT) images. However, hemorrhagic lesions can be complex and difficult to distinguish manually. To solve these problems, here we propose a novel Hybrid 2D/3D UNet deep-learning framework for automatic aSAH identification and quantification in NCCT images. We evaluated 1824 consecutive patients admitted with aSAH to four hospitals in China between June 2018 and May 2022. Accuracy and precision, Dice scores and intersection over union (IoU), and interclass correlation coefficients (ICC) were calculated to assess model performance, segmentation performance, and correlations between automatic and manual segmentation, respectively. A total of 1355 patients with aSAH were enrolled: 931, 101, 179, and 144 in four datasets, of whom 326 were scanned with Siemens, 640 with Philips, and 389 with GE Medical Systems scanners. Our proposed deep-learning method accurately identified (accuracies 0.993-0.999) and segmented (Dice scores 0.550-0.897) hemorrhage in both the internal and external datasets, even combinations of hemorrhage subtypes. We further developed a convenient AI-assisted platform based on our algorithm to assist clinical workflows, whose performance was comparable to manual measurements by experienced neurosurgeons (ICCs 0.815-0.957) but with greater efficiency and reduced cost. While this tool has not yet been prospectively tested in clinical practice, our innovative hybrid network algorithm and platform can accurately identify and quantify aSAH, paving the way for fast and cheap NCCT interpretation and a reliable AI-based approach to expedite clinical decision-making for aSAH patients.

YNICL Journal 2019 Journal Article

Severe asymptomatic carotid stenosis is associated with robust reductions in homotopic functional connectivity

• Lei Gao
• Tao Wang
• Tianyi Qian
• Feng Xiao
• Lijun Bai
• Junjian Zhang
• Haibo Xu

Severe (>70% narrowing) asymptomatic carotid stenosis (SACS) is associated with cognitive impairment and future strokes, and connectivity basis for the remote brain consequences is poorly understood. Here we explored homotopic connectivity and parenchymal lesions measured by multimodal magnetic resonance imaging (MRI) parameters in patients with SACS. Twenty-four patients with SACS (19 males/5 females; 64.25 ± 7.18 years), 24 comorbidities-matched controls (19 males/5 females; 67.16 ± 6.10 years), and an independent sample of elderly healthy controls (39 females/45 males; 57.92 ± 4.94 years) were included. Homotopic functional connectivity (FC) of resting-state functional MRI and structural connectivity (SC) of deterministic tractography were assessed. Arterial spin labeling based cerebral perfusion, susceptibility weighted imaging based microhemorrhagic lesions, and T2-weighted white matter hyperintensities were also quantified. Significant and robust homotopic reductions (validated by the independent dataset and support vector machine-based machine learning) were identified in the Perisylvian fissure in patients with SACS (false discovery rate corrected, voxel p < 0.05). These involved regions span across several large-scale brain systems, which include the somatomotor, salience, dorsal attention, and orbitofrontal-limbic networks. This significantly reduced homotopic FC can be partially explained by the corrected white matter hyperintensity size. Further association analyses suggest that the decreased homotopic FC in these brain regions is most closely associated with delayed memory recall, sensorimotor processing, and other simple cognitive functions. Together, these results suggest that SACS predominately affects the lower-order brain systems, while higher-order systems, especially the topographies of default mode network, are least impacted initially, but may serve as a hallmark precursor to vascular dementia. Thus, assessment of homotopic FC may provide a means of noninvasively tracking the progression of downstream brain damage following asymptomatic carotid stenosis.

YNIMG Journal 1998 Journal Article

Attention-Sensitive ERPs Elicited by Kinetic Forms

• Jiongjiong Wang
• Yinpeng Jin
• Feng Xiao
• Lin Chen